<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Daniel Wu on Medium]]></title>
        <description><![CDATA[Stories by Daniel Wu on Medium]]></description>
        <link>https://medium.com/@danielwu3?source=rss-a2bc7fe3d2d4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*DPfFDbVV0QGhjFFDOkvwxg.jpeg</url>
            <title>Stories by Daniel Wu on Medium</title>
            <link>https://medium.com/@danielwu3?source=rss-a2bc7fe3d2d4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 07:18:13 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@danielwu3/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Recap: Future of Care Delivery Summit at Health Tech Austin (Feb 2023)]]></title>
            <link>https://medium.com/@danielwu3/recap-future-of-care-delivery-summit-at-health-tech-austin-feb-2023-8bc0712c9ad7?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/8bc0712c9ad7</guid>
            <category><![CDATA[health-tech-trends]]></category>
            <category><![CDATA[healthcare]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[health-technology]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Tue, 21 Mar 2023 17:38:58 GMT</pubDate>
            <atom:updated>2023-03-21T17:38:58.114Z</atom:updated>
            <content:encoded><![CDATA[<p>On February 23, 2023, Health Tech Austin hosted the “Future of Care Delivery Summit.” This was a half day event at the Capital Factory in downtown Austin. Health Tech Austin is an organization that hosts healthcare related events, bringing together healthcare subject matter experts to the Austin community. Capital Factory is Austin’s central hub for the tech and startup community based in the Omni building in downtown Austin.</p><p>The speakers for the event included. Summaries of the sessions follow below with closing thoughts at the end.</p><ul><li>Desmar Walkes, MD — Medical Director at Austin Public Health</li><li>Hugh Forrest — Co-president at SXSW</li><li>Fred Turner — Co-founder/CEO at Curative Health</li><li>Mike Geeslin — President/CEO at Central Health</li><li>Kacie Kelley — Chief Innovation Officer at Meadows Institute</li></ul><p>In the opening Dr. Desmar Walkes, medical director at the City of Austin, calls on the audience, “what light can we shine on issues of inequity and bias?” I agree that technology can play a role (e.g., telehealth), outcomes came from meeting patients where they are, and care delivery is a partnership with technology. There’s so much that can be said out of these few points. My personal perspective is that technology must play a supporting and enabling role. It’s rarely the case it’s the central character.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/200/1*nlVrfUInMt5qsGMpE6tOPw.jpeg" /><figcaption>Dr. Desmar Walkes. Austin Public Health, speaks at Health Tech Austin — Feb 2023 (Source: <a href="https://www.austintexas.gov/news/city-austin-announces-new-medical-directorhealth-authority">https://www.austintexas.gov/news/city-austin-announces-new-medical-directorhealth-authority</a>)</figcaption></figure><p>Further, Dr. Walkes highlighted health inequities and social determinants of health in the Austin area. In her role, she is the health authority for public health for the Austin metropolitan area, working with community leaders and elected officials. In Austin, 25% of black and 22% of hispanic population are unable to receive medical care. Her organization identifies gaps in resources and works to bring those needed resources to the population at need.</p><p>In a later session, Mike Geeslin, President and CEO of Central Health, which is Travis County Healthcare District, pointed out that their focus is on the Travis County population that is 200% below the poverty line. That amounts to 7.8% of the county’s population, which is 23K people below the federal poverty line. The focus of the organization is through partnerships, contracts, and even hiring clinicians directly to make improvements as measured by an index. Not surprisingly, Austin is a key access point for many for healthcare.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0wHBlXELYuF8TpA7isrnqQ.jpeg" /><figcaption>Mike Geeslin, CEO of Central Health, speaking at the Health Tech Austin — Feb 2023</figcaption></figure><blockquote>Health inequities are differences in health status or in the distribution of health resources between different population groups, arising from the social conditions in which people are born, grow, live, work and age. Health inequities are unfair and could be reduced by the right mix of government policies. (<a href="https://www.who.int/news-room/facts-in-pictures/detail/health-inequities-and-their-causes">World Health Organization</a>)</blockquote><p>Health inequities can be surfaced by looking at social determinants of health (SDoH). US Department of Health &amp; Human Services depicts the following five domains:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/910/1*BSPe_3HgIp-doVb5xpT4zw.png" /><figcaption>Source: Healthy People 2030, U.S. Department of Health and Human Services, Office of Disease Prevention and Health Promotion. Retrieved [date graphic was accessed], from <a href="https://health.gov/healthypeople/objectives-and-data/social-determinants-health">https://health.gov/healthypeople/objectives-and-data/social-determinants-health</a></figcaption></figure><p><strong>Fred Turner, Co-founder/CEO of Curative Health</strong>, shared history on his journey in fixing healthcare, and why Curative was focused on a new $0 out of pocket medical plan. Curative was founded in January 2020 working on the sepsis problem. Through the COVID shut down, the company pivoted to lab diagnostics. The team rapidly grew to thousands of employees, built labs in Austin and DC, and had 19,000 testing sites in the US. Through the pandemic, Curative worked with payers saw how sub-optimal the billing experience was. Free COVID tests were being charged by insurers. In the lab testing space, what could be done? Either improve the codes used for billing. If there was a new diagnostic, there was no code and it required getting a code in order to be reimbursed.</p><p>Out of this experience, Fred led Curative to look at a new approach to make healthcare better for patients. They looked at the high deductible health plans (HDHP), which is a very popular type of plan. When looking at the data, the HDHP didn’t really improve patient outcomes. Research on 15k members over 3 years showed that costs going down in Year 1 below a baseline, meeting baseline costs in Year 2, and exceeding baseline in Year 3. Costs went down 12% because people were avoiding preventative visits e.g., colonoscopies. Due to the delayed preventative care, it amounted to increased costs e.g., colon cancer later on.</p><p>As result, Curative is rolling out the Curative Health Plan, starting in Austin TX and then rest of Texas in 2023. More information can be found on its website. Essentially, the idea is to lower the cost by having no co-pays or deductibles. This removes the financial and psychological barrier from going for preventative care. From Curative’s perspective, every dollar spent on preventative care reduces future costs. The goal is to reduce cost for preventative care and increase access to preventative care.</p><p>The plan incentivizes preventative care in the first 120 days for its subscribers. Get the preventative care and the respective tests done, and you get to stay on this no co-pay and no deductible plan. Otherwise, the subscriber deals with a deductible. It includes spending time with a care navigator to help the person understand where to get care and when to get care at a hospital. It includes time with clinicians as well to identify gaps in care ahead of time, and link the member with primary care provider.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s1xFDybisQc62GGUYFSjog.jpeg" /><figcaption>Clay Johnston, co-founder of Harbor Health speaking at Health Tech Austin — Feb 2023</figcaption></figure><p><strong>Clay Johnston, Co-founder/CMO of Harbor Health, </strong>shared his career arc from academia to healthcare start up. Dr. Clay Johnston was the former Dean of the Dell Medical School at UT Austin among many distinguished academic posts. In his session, he shared about a story of his days in neurology where treatment to a specific stoke condition was the result of existing practice beliefs as opposed to evidence. Commercialization from research is a very slow process so he sought out to fix the research system. Paraphrased, his conclusion was that “the interests of the system was disconnected with making improvements and won’t solve patient outcomes.” Millions of dollars were spent on drugs at the expense of neglecting care delivery and preventative care. His goal is to redesign the care system and reduce the waste being spent.</p><p>I sought out to add more details in the session, which was limited by time. In Joe Lonsdale’s podcast “American Optimist,” Jon interviews Clay in <a href="https://podcasts.apple.com/us/podcast/joe-lonsdale-american-optimist/id1573141757?i=1000598711092">episode 44</a>. Clay elaborates on the challenges of the healthcare system — the speed of innovation, areas of focus, and integrating into the care delivery. Below are my summarized highlights from the podcast.</p><p><strong>Why is healthcare so broken? </strong>Doctors respond to illness in a visit in a hospital or clinic but that’s a small part of healthcare. Once patient gets seen and patients get sent out, only half of patients take their prescriptions as prescribed 3 months after. Healthcare doesn’t address affordability, accessibility, or rewards adherence and follow up. This is due to how healthcare is paid for. It pays for writing the script but not more. Healthcare systems are limited, meaning our clinicians don’t know where to send patients to get better care. Productivity from healthcare provider perspective is getting more revenue.</p><p><strong>Is value-based care and applying technology working</strong>? It’s working better but not quite so bold. The incentives for the patient, provider and longer term for patients need to be aligned. Medicare Advantage is shifting the incentives a little but there’s still more room for innovation.</p><p><strong>What should the redesign of the healthcare system look like</strong>? Patients reported outcomes. Is it superb? Avoiding things that don’t need to be done. Focusing on things that don’t reimburse well in fee-for-service that need to be done. For example, in knee pain, it’s weight loss, PT, training, etc. to avoid unncessary surgeries. They’ve put doctors on salaries.</p><p><strong>How are the payors incentives misaligned</strong>? Joe describes that payors would think that they would want to pay less. However, the capping of the profits means that the Medical loss ratio (MLR) has to be 80% or higher. Margins are set at 20% and if spend is less than 20%, payors are losing money.</p><p><strong>Could self-insured employers do this better? </strong>Employers pushing back to payor and payor didn’t have the incentives.</p><p>Patients don’t know about quality and there was no marketplace. Medical school was used to prove that a realigned incentive can work, but it wasn’t scalable. Harbor Health is the redesigned system — redesigned insurance, primary care navigation, population health tools, etc. Can pay for and direct patients to the best specialists to get the best care.</p><p>Tony Miller is the co-founder at Harbor Health and came from the health insurance side. From visibility into healthcare data, high quality care usually don’t have high costs. Harbor Health is in Austin with 3 clinics with team based care, and mobile clinics. Offering insurance plan via employers as a starting point, and Medicare.</p><p><strong>Kacie Kelly is the Chief Innovation Officer of </strong><a href="https://mmhpi.org/"><strong>Meadows Institute</strong></a>, which focuses on mental health policy in Texas. I found this session particularly interesting as I was not close to the health policy organizations on a regional level. The Meadows Institute goal is to be a national leader in mental health including the treatment of all people. The institute works on testing solutions with a range of “assets” at scale. While we often hear of health tech startups addressing an aspect of need in healthcare, I get the sense that Meadows Institute is looking for direct, scalable impact with tools or capabilities available today.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/534/1*NCqbTYk3ZoYWGxXeshVXqg.jpeg" /><figcaption>Kacie Kelly, Chief Innovation Officer at the Meadows Institute, speaks at Health Tech Austin — Feb 2023 (Source: <a href="https://mmhpi.org/about/our-team/">https://mmhpi.org/about/our-team/</a>)</figcaption></figure><p><strong>Background on mental health in Texas</strong>: In 2021, 4,000 lives in Texas died of suicide and 11,000 lives affected by substance abuse</p><blockquote>“The mental health system is not broken, it’s working as designed” — Kacie Kelly</blockquote><p><strong>The problem</strong>: Mental health is addressed outside of the healthcare system. It’s diagnosed in the ER or in the justice system. It’s too late. When the mental health needs were addressed earlier, it’s by a specialist regardless whether it needs a specialist. Bottom line is that this is a bottle neck in resources and is too heavily dependent on the limited specialist workforce.</p><p>Result of this issue is that there are poor outcomes. Kacie shared interesting data points on the limited percentage of adults who get care for depression. With adults in the US, 20 million adults have depression and yet less than half get treatment and care. Of that number 25,000 get better. With youth in the US, 50% of the mental health issues start at 14 years old and 70% by 26 years old. “It’s critical to offset the trajectory.”</p><p><strong>The challenges</strong>: It takes about 11 years from first sign of symptoms to treatment. Think about the access to care considerations. In light of recent COVID impacts, 250,000 youth lost a primary care giver due to COVID. Additionally, how do the schools integrate into the healthcare system.</p><blockquote>The mental health system needs a “paradigm shift” to detect early and treat starting in schools. — Kacie Kelly</blockquote><p>With these challenges, their focus is on advancing today’s solutions and increasing access. Here are two examples how the Meadows Institute is tackling the challenges:</p><ol><li>Working with health systems on behavioral health initiatives in the primary care and pediatric care setting. At John Hannah (sic), the work is on integrating behavioral health through changes in workflow and training. There’s collaborative care to ramp up delivery. Initiative to ensure support for behavioral health care is financially sustainable and covered by Medicare/Medicaid. The goal is to reach 6 million Texans by end of 2023, 400,000 lives get treatment, and 150,000 lives have readmission. In 2026, the goal is to reach 10 million Texans, 3 million get treatment, and 1 million have readmission. In the longer term, the work expands to initiatives with the workforce and other areas such as embedding digital solutions to enable this work.</li><li>Linking schools to mental health systems. The <a href="https://tcmhcc.utsystem.edu/">Texas Child Mental Health Consortium</a> connects healthcare systems to mental health care with youth. There’s telehealth, child psychiatry, and other initiatives.</li><li>Collaborations with the innovation community. There’s the launch of the mental health ventures to accelerate and catalyze the generation of ideas and solutions.</li></ol><p>In closing, the Health Tech Austin event in February was full of subject matter experts and leaders in the healthcare space. Each could have been an area of specific focus.</p><p>Social determinants of health and health equity is an important focal point that will continue to drive innovators to ask how technology, programs, and processes can improve equity. How can the local organizations, existing or new plays contribute to improvements? How can incentives be aligned along this level to improve equity?</p><p>Within healthcare, the health tech space is dynamic and full of challenges. Having been in health tech for more than five years, exposure to other innovators and providers in this space was refreshing. To me, it’s seeing and acknowledging the essential role that local, community, and non-profit organizations play in the delivery of care. Technology plays an enabling and supporting role, rarely brings value by playing a central role.</p><p>While there are systemic challenges, organizations in the health policy space can make an impact by focusing on today’s solutions and broadening the delivery of care through existing pathways. It makes me wonder whether tech has enabled health policy and non-profit organizations optimally. In other cases, working outside of the system was the preferred approach as in the stories that Fred with Curative and Clay at Harbor had shared.</p><p>There’s no one unified approach since the surface area is so broad. There’s a sense of the shared mission, and different players in the healthcare ecosystem working to improve the system, improve the care delivery, and improve patient outcomes.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8bc0712c9ad7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Traveling through my earliest memories of a childhood in Taiwan]]></title>
            <link>https://medium.com/@danielwu3/traveling-through-my-earliest-memories-of-a-childhood-in-taiwan-2e1574dab15?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/2e1574dab15</guid>
            <category><![CDATA[asian-american]]></category>
            <category><![CDATA[childhood]]></category>
            <category><![CDATA[memories]]></category>
            <category><![CDATA[writehere]]></category>
            <category><![CDATA[taiwan]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Sun, 22 May 2022 15:42:20 GMT</pubDate>
            <atom:updated>2022-05-23T03:21:06.591Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Image of a mysterious place shrouded in fog" src="https://cdn-images-1.medium.com/max/1024/0*e5jWmMOk9yGyNPez" /><figcaption>Photo by <a href="https://unsplash.com/@philippeyuan?utm_source=medium&amp;utm_medium=referral">Philippe Yuan</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>There’s a memory of a time long ago that lives in a distant pocket of my mind. It’s hidden behind years of experiences, weathered by the meanderings of life. If there was such a thing as time and space travel, this would be it. I’m talking about another kind of travel. The kind it takes to search deep inside — past the voices of today, the 4k images, and the visceral swirling feelings of the present — clouded and faded by the distance.</p><p>Here, I’m talking about the memories of childhood. This is of a particular time in my life. This travel brings me to a place in Taipei, Taiwan. Not with the modern skyscrapers, but of another time, another place. This is of a youth in my childhood. At a corner in the Jingmei district, street life buzzed with cars, buses, and trucks stopping at the lights of this busy intersection. That world — outside the concrete walls and beyond this garden of sorts — mattered little to me.</p><p>Within these walls, there was a lush, verdant garden. Tropical is what I see. Within this garden sat a cube-like concrete gray home. That was my home to my Gong gong and Pua pua. In the living room floor, I sat facing a wood-cased TV console in the corner, watching my favorite show. The images of Tsu Liu Shang (maybe spelled Chu Liuxiang) played. It was a TV-series about a historical Chinese hero who had various adventures, helping people along the journeys of his life. The show was a drama set in pre-modern times. He wore a traditional silk gown and held a hand fan with a way that was his style, distinct to his character.</p><p>To my right, traditional, carved rosewood furniture was the right side of my back. Its pillows were rectangular, properly flat, firm, and nicely wrapped in a plastic that it said to “don’t sit here.” I can’t recall anybody really sitting there, at least for kids.</p><p>Further behind me, there was this dining room in the same style of decor. Large table and chairs, seating for eight. An empty table but with an occassional dish readied under a plastic fly cover. I hear noise from the hallway leading to the kitchen, left of the dining room. Grandma, who I called Pua pua, was in the small kitchen preparing the next meal. This kitchen being tight was not a particularly welcoming place for a young child even though the smells of the food was alluring. Teal tiles lined the walls from floor to wall. The clanging of the pots and pans rang, banging and cooking on a propane stove. I smelled the burning of the gas stove on top of the food. How Pua-pua was able to cook meals in such a small, dimly lit kitchen for the family, I don’t know.</p><p>I hear the slam of the front door and the sound of someone coming home. Maybe it was Hsiao Jio-jio, my uncle (Literally, “little uncle” who was my mom’s youngest brother). Maybe, it was Gong gong, my grandpa, returning from work. From the living room on the right past the furniture, I look over at the hallway inside of the front door. It’s a rather dim space for the day time. Shoes lined up inside the hallway. This same place was where my mom and her siblings grew up. I suppose their shoes filled the floor, maybe chaotically. I suppose perhaps not. I say that because Gong gong was a former military officer whose disciplinarian standards carried over into family life — or so I was told by my mom.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*QNIUDAz5clCbEqJyc8aLfQ.jpeg" /><figcaption>A dark hallway and stairs leading up at my grandparent’s house</figcaption></figure><p>The moving images from the TV show briefly caught my attention. A male vocal sang with instrumentals, playing the theme song faded. Dusk was falling through the large glass sliding doors on my left. It definitely had a view. During the day, ample daylight shone through the glass sliding doors. The view through the yard was limited. It was hard to call it a backyard. It was more like a lush green estate that abruptly met the wall of a rising hillside. At the end of this property, the face of the hill looked back at the home and the windows of the sliding doors where I sat. This face impressed upon me to be a mountain side.</p><p>I walk to the stairway near the front entrance inside the home. At the foot of the stairway, I start the steps on my way up and around to the bedrooms on the second floor. Bedrooms line the left and right. The two sides of the floor plan greet me. “Don’t go to your Gong gong’s room,” grandma reminded me as I look over to the left. Why couldn’t I go into his room, I thought? What was in there? It seemed like it was his personal study and bedroom. The door was typically shut closed, and rarely have I ever had an opportunity to peek in.</p><p>On other occassions, it was the day of the week that grandma did the laundry. From the second floor, a stairway led up to the roof top. Grandma would go up to hang the laundry outside above the home. She went up through the stairs to the roof which I was not allowed to go. The door to the roof cracked open with sunlight peering down.</p><p>As it turned late now, I turned around to go to bed in grandma’s room on the right from the stairway. As I laid in bed, Pua pua would lay there next to me in her own bed in the same cozy room. It was a balmy evening as usual. I always wondered what she was thinking about. I knew she was not yet asleep, awake with quiet thoughts in the evening air. She had a habit picking at the skin around her fingernails — a habit which I find myself doing when bored. I am reminded of her when I catch myself doing that too. It was dark in the room but I could make out her silhouette in the moonlight. Beyond the bedroom window, there was the presence of the night sky and the moonlight entering. At the top of the mountain side, some business peered over slightly. Perhaps, it was an inn. The glow of the neon sign was unmistakabl, glowing in the distance where it drifted into the night sky. The moonlight came and went with the evening.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M4zUcJs2S-YZjMz_bPiZGA.jpeg" /><figcaption>A view of Pua pua sleeping next to me in her bed, evening lights through the window</figcaption></figure><p>I had ample opportunities to play. I had a palm-sized pet turtle who lived in a plastic shallow aquarium. I kept my turtle there inside the house. I had my bicycle that I rode around the outside patio. I remember riding in Taipei only to forget and then relearn to ride later in Houston. I played in the patio area outside the front entrance. I played in the back yard — yes the one that met the face of a mountainside.</p><p>Sometimes my uncle Hsiao Jio-Jio was around. He was a younger guy then. Small build but not slight. An earnest face with glasses. A gentle voice and even tempered. “Hsiao Jio-Jio!” I would call him. “ Look, I found something in the grass!” He played along and helped to carry the largest snail I’ve ever encountered. We placed it into the white ceramic sink next to the front door, and continued playing.</p><p>There was a world outside these concrete walls around the property. One day, Pua pua and I ventured out to the market for groceries. We browsed and bought what we needed. We carried what we could by hand, and we had a portable wheeled cart. We bought rice this time. The man at the grocery said, “Good job young man. Help your grandma.” So I did, being helpful. The sack of rice was on my shoulders as I carried it home for grandma. It was heavy.</p><p>On Sundays, we went to church. We took a bus with the words “MCI” on the front. It picked us up on the corner of the bustling street outside the walls. The bus arrived with a roar and stopped with a grumble and hiss. People got off. Time to get on. Grandma went first and took my hand. I was right behind. It was the most intimidating thing to get on and off. One foot off of the ground, over the gap, and on to the step of the bus at the door. Other foot in the bus. Climb up the steep steps and into the bus. Would the bus drive off right as I was stepping off of the curb into the bus? The first step is always to biggest. It’s true for life too. It seemed like a long ride and a quite a journey with walking. I don’t remember what the pastor spoke about. We sat with a small group of adults at the church.</p><p>Back at the home, Gong gong seemed to be busy coming and going everyday. He had a driver who would drive him to work and back home. The gate would open and his car would enter. The car would be parked in the port right of the gate entrance. At the end of the day, the driver would put on his helmet and get on his motorcycle to return home. For years as a child, I associated that driver with the face of a specific family friend in a different place. It couldn’t be the same person but I wanted to believe that he was associated with my grandpa. Did the driver have such pronounced eyebrows like this family friend? It couldn’t be, but there was a resemblance in my mind as a child. I wanted to believe.</p><p>That home — the place and the people — has a special place in my childhood. Certainly, this home was where my mom, my Ah-yi’s, and Jio jio’s grew up. Imagine them as children themselves. The busy intersection. The concrete solid fence. Nestled at the base of a mountainside. Perhaps in their time, it was the bustling place in my memories. The routines of life room to room passed through the decades and generations. Through school to growing up and leaving their home for America. Until, the home stood forgotten for years in a modern, growing city. It gave way to progress — a road expansion.</p><p>This home only exists in my memory now. I am far from that place, in time and place. Memories rely on context — a physical item, a smell, a feeling, sounds and people. That life was a continent away. That was real, yet it doesn’t quite feel any different from another distant thought in my head. It lives in my fading texture of my day to day busy life. But memory is interesting. It starts like a first person video, and then fades into a third person … and piece by piece it slowly disappears. Literally, an old penny wearing out worn with time. Next time, what will be left should it not be visited?</p><p>Those were mainly my memories when I was five years old, peppered with a visit before I was ten. Those were memories of almost a year in Taiwan. When I meet other friends and acquaintances come from Taiwan, it’s hard for me to say that my experience was anywhere like theirs. Mine was dream-like but real. Certainly, theirs was real too with schooling and friends. For me, this part of me was disentangled from the rest of my childhood and so distant from who I am even today. Still, these share these treasured memories of mine like an old dusty memorabilia sitting on the shelf all by itself.</p><p>Thank you for sharing your time to walk with me. Although my grandparent’s home doesn’t exist anymore, it would be interesting to find out the location of the home in the Jingmei district. There was a police station nearby. There was a tunnel road. Where could it be?</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.google.com%2Fmaps%2Fembed%2Fv1%2Fplace%3Fcenter%3D24.9940679%252C121.5331906%26key%3DAIzaSyD9HrlRuI1Ani0-MTZ7pvzxwxi4pgW0BCY%26zoom%3D15%26q%3DJIngmei%2BNIght%2BMarket%2C%2BWenshan%2BDistrict%2C%2BTaipei%2BCity%2C%2BTaiwan%2B116&amp;display_name=Google+Maps&amp;url=https%3A%2F%2Fwww.google.com%2Fmaps%2Fplace%2FJIngmei%2BNIght%2BMarket%2C%2BWenshan%2BDistrict%2C%2BTaipei%2BCity%2C%2BTaiwan%2B116%2F%4024.9940679%2C121.5331906%2C15z%2Fdata%3D%213m1%214b1%214m5%213m4%211s0x3442aa015a6bb427%3A0x554e4f5f7a6d4ada%218m2%213d24.9940492%214d121.5419454%3Fshorturl%3D1&amp;image=http%3A%2F%2Fmaps-api-ssl.google.com%2Fmaps%2Fapi%2Fstaticmap%3Fcenter%3D24.9940679%2C121.5331906%26zoom%3D15%26size%3D250x250%26sensor%3Dfalse&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=google" width="600" height="450" frameborder="0" scrolling="no"><a href="https://medium.com/media/c911c8d10afa3d746d4285ca628cced5/href">https://medium.com/media/c911c8d10afa3d746d4285ca628cced5/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2e1574dab15" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How my product reached 1,000,000 users]]></title>
            <link>https://medium.com/@danielwu3/how-my-product-reached-1-000-000-users-b4cf3ac385b7?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/b4cf3ac385b7</guid>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[product]]></category>
            <category><![CDATA[product-strategy]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[growth-hacking]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Mon, 10 Jan 2022 19:26:46 GMT</pubDate>
            <atom:updated>2022-01-10T19:26:46.517Z</atom:updated>
            <content:encoded><![CDATA[<h4>The customer journey approach that product management leaders should use to drive growth</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*J5pVp3LSD_6gcPQF" /><figcaption>Photo by <a href="https://unsplash.com/@jeremythomasphoto?utm_source=medium&amp;utm_medium=referral">Jeremy Thomas</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Previously, we discussed how product success does not stop at general availability (aka “GA”). Rather, product success must be measured by customer adoption. We discussed why that is the target and why that is important. We touched on some of the key metrics to measure adoption by. Read more <a href="https://medium.com/@danielwu3/product-success-does-not-stop-at-general-availability-ga-5c417380e1ac">here</a> for more background on this essential aspect of product management.</p><p>In this post, I want to cover the “how” of doing product client adoption at scale. For product leaders, this covers product strategy, vision, and organizational design. I draw these lessons based on my experiences in product at a recent large health company and a very large enterprise tech company.</p><p>The key is to think about this from the start and build it into your product priorities so it’s seamless as you move each incremental customer through the adoption journey. Just as a SaaS product should not support several instances of an application (e.g., one for each client), there’s an economies of scale dynamic here with adoption. The idea behind adoption is that resources will be limited and expect constraints. As result, the adoption approach must be a systems approach, functioning whether you’ve one client or hundreds of clients.</p><p>The goal is to identify the key areas required to frame your customer adoption system. It’s not meant to be in depth or comprehensive. As we go through these areas, it may require building up your team. It may require work on the organization level with leadership across your company to define processes and programs. It’s a process and journey, but it’s worthwhile and rewarding.</p><h4>Evaluate and Assess</h4><ul><li><strong>Map the Customer Journey</strong>: The journey is a map of the client’s touch points with your product (or company’s) across the lifecycle from customer acquisition to retention. This is an opportunity to pause and ask whether this existing or default journey fits the missions and outcomes of the company at this point. Start broad and early in the journey. Who should articulate the customer journey? Be collaborative, including marketing, sales, design, engineering, and support — especially when this is not well understood or time/context has changed. By building the customer journey together, this an opportunity to have a shared vision across all the stakeholders building towards a specific adoption outcome. Often times, the output can be not just one journey but multiple journeys. Does this mean we should have only one journey? The answer is that it depends on the business and the goals.</li><li><strong>Measure the Current State Metrics</strong>: Based off of the journey mapping, quantify the current state through metrics. This provides a data-centric view of the journey as-is. It provides an opportunity to see X% moving from stage A to B, and Y% from stage B to C, and etc. These are the levers toward progress. If you’ve a purely digital journey, then having analytics data (instrumentation) is essential. Work with analytics or engineering provide support to you or the product team to query those numbers. It can be done ad hoc or manually at this point. These are reference target metrics from which we can optimize in the journey.</li><li><strong>Uncover Silos in Customer Data</strong>: Through the effort of evaluating/assessing metrics, expect to encounter situations where the data lies in separate data siloes. Connect customer data silos and lay the foundation for automation of insights. That way you don’t need to rely on ad hoc requests to product analytics or engineering, which is costly for your organization. For example, what is the percentage of customers that utilized a specific workflow when the numerator count is on the product application in Amplitude and the denominator count is on a Salesforce customer portal? Some data could live in Salesforce and others in a data warehouse like Snowflake. Having connected tools and infrastructure can help streamline the awareness on the state of adoption. Build those organizational bridges and agreements so the product managers won’t be left fighting uphill battles.</li></ul><h4>Define the Target Adoption Metric and Build the Future State</h4><p>Build a vision of what the preferred state should be. It may be prioritizing what can be achieved in the near term (e.g., clear CTAs to sign up for a beta in Success Community) over longer term constraints (e.g., Learning Portal accessibility). Build the preferred customer journey map. Define the primary adoption target metric and the supporting metrics at each stage. Include stakeholders across disciplines, and include the champions. When there are multiple product lines that are related, consider the opportunities to upsell and cross among the products.</p><h4>Execute through Experiments</h4><p>Like building a product or feature, there are MVPs, Version 1, Version 2, and so forth. Build the offering and the client journey as the product. This system thinking based on the customer journey is a macro-level scale from the product-level user journey. This is where we want to learn quickly by tackling high value, lower effort opportunities. Build towards the vision. Demonstrate small wins. Tell the story to build buy in among key stakeholders. Gather customer insights quickly and observe the outcome. Repeat over and over until we see movement towards the adoption goal.</p><p>There’s a term called growth hacking that is popular that originated in the startup world. It’s really a mindset and philosophy that can be implemented in the organization. Whether your organization is an early stage startup or not, I think even established companies and organizations can learn a lot about the strategies that can power growth. My favorite author and go to on the subject is Sean Ellis who wrote “<em>Hacking Growth</em>” (Note: I receive commissions from purchases from below affiliate link).</p><p><a href="https://amzn.to/3GbckGK">Hacking Growth: How Today&#39;s Fastest-Growing Companies Drive Breakout Success</a></p><h4>Build Growth and Supporting Organizational Functions</h4><p>Longer term, sustaining continued growth requires a cultural change in the organization, not only in product. It’s in the best interest of a product leader to put into place an organization that empowers product growth. The growth function is a team sport that cuts across disciplines beyond marketing, sales, support, or product alone. That may not be the case today, and requires building the vision and selling the vision to respective leaders. It’s beyond messaging and positioning in marketing. It’s beyond analytics function alone to execute on the vision. It’s beyond product function alone since the product should support the offering and the customer journey. There’s a few common themes to drive adoption and growth:</p><ol><li><strong>Product education experiences</strong> — Collaboration between product and content that is either in app or more seamless. Result: Allows better customer retention.</li><li><strong>Customer support experiences</strong> — Collaboration between product and customer success, technical support, and related. Result: Allows better customer retention.</li><li><strong>Trial experiences</strong> — Collaboration between product and marketing, customer success or sales. This can be transformational in repositioning product has a revenue generator. Result: Allows customer acquisition.</li><li><strong>Purchase experiences</strong> — Collaboration between product and finance. This can be transformational in repositioning product has a revenue generator. Result: Allows customer acquisition.</li></ol><p>I appreciate any feedback on what topics you would like to read. If you enjoyed reading, please give this some claps below. Click on the Follow to subscribe to future posts.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b4cf3ac385b7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Product success does not stop at General Availability (“GA”)]]></title>
            <link>https://medium.com/@danielwu3/product-success-does-not-stop-at-general-availability-ga-5c417380e1ac?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/5c417380e1ac</guid>
            <category><![CDATA[product-manager]]></category>
            <category><![CDATA[growth-hacking]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[product-strategy]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Wed, 15 Dec 2021 16:13:00 GMT</pubDate>
            <atom:updated>2021-12-15T16:13:00.359Z</atom:updated>
            <content:encoded><![CDATA[<h4>GA is only one stop along the way to success</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*KdRTEEJypRNTqWxu" /><figcaption>Photo by <a href="https://unsplash.com/@purzlbaum?utm_source=medium&amp;utm_medium=referral">Claudio Schwarz</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>I applaud all the product managers and product teams that have taken a product from an idea into any further stage — alphas, betas, or General Availability. It is no small feat. I speak to the product managers and product leaders, particularly to those in <em>large B2B technology firms that have large product teams supporting a multitude of products and offerings</em>. The message is this. Delivering a product to General Availability should not be conflated with success.</p><blockquote>The ultimate product success is measured by client adoption, not only by delivery.</blockquote><h3>Success is Client Adoption</h3><p>Making products reach General Availability is worthy of celebration. I’ve walked in your shoes. BUT — It isn’t sufficient of a goal in it of itself. Rather, it is sufficient once clients have adopted your product. Then, the value of the product becomes tangibly real between the client’s users and the client, and between the client and the product. I want to share my thoughts why the larger goal of client adoption should be central to the product manager’s definition of success.</p><p><strong>Why is the focus on client adoption of a product not obvious?</strong> In theory, any product manager will say it’s obvious. The reality of the job means there are many competing interests and distractions. Adoption can easily be forgotten, or important only in name. It’s time to re-focus and value to the client, and the top reasons include:</p><ol><li><strong>Client retention and satisfaction</strong>: Client adoption means client retention and satisfaction, which minimizes churn. When clients demonstrate strong adoption, then it’s an indicator of their satisfaction, finding utility and usefulness in a particular product.</li><li><strong>Key to new client acquisition</strong>: When the business context drives the need for new client acquisition, it is measured by adoption. A client may require a pilot or trial, which is an initial adoption. Hurdles exist to challenge full adoption.</li><li><strong>Improves client journey and ROI</strong>: The client journey is a means to achieve adoption. The entire client journey becomes the product. These journeys will reveal opportunities to accelerate adoption, which has a flywheel effect on client satisfaction. When done right, this can be a significant competitive differentiation.</li></ol><p><strong>What is considered to be reaching the “client adoption” of a product?</strong> Client adoption of a product is a realized outcome over a measurable target. One common measure is a period of time. This outcome is the threshold where the client chooses a specific product in addition to OR over the other options in this period of time. It’s the result of a continuous process to move clients from one stage of the client journey to the next. Let’s define this. The outcomes can be measured by Y over the course of Z periods. Here are a couple examples:</p><ol><li><strong>Initial Adoption</strong>: Example — Percentage of clients who used the product (e.g., scheduling app) over the total clients who are eligible within the past year. This is a lower bar to clear, and measures the client interest. Clients get their foot in the door. It does not measure stickiness. This measurement has to be paired with a more long term measure such as the following.</li><li><strong>Portfolio Adoption: </strong>Among all options the client may use, what percentage is your product in the mix of all the options the client uses? This perspective is akin to a share of wallet mindset, similar to social media channels. The process to grow into a meaning percentage is essential. That is the bar to measure up to, and is the longer term goal.</li></ol><p>There are number of ways to define success, and I realize every organization and situation varies. These adoption goals should be part of the Objective-Key Result for your product or organization. Those are two high level examples, and underneath these macros metrics, adoption is made up of a series of tactical measures tied to the client journey. I will cover this in more detail in a future article.</p><p>Caution for those with tendency to be forgetful. Adoption is not a one time activity that you set and forget. It’s not a light switch that turns on and off. Similarly, client churn is not on and off either. In many cases, a signal that clearly states the status of client adoption being absolutely 100% yes is not feasible. If so, it’s usually too late when we have that signal that it’s a 100% not adopt (or churn). Clients can start adoption — not reaching the portfolio adoption stage — only to churn.</p><p>Finally, I want to speak on organizational support for adoption. Since client adoption is not a distinct one-time activity in the product delivery, it’s not a checklist before product planning. It’s not just a specific point in time after GA either. If it is, it’s also not going to mature and yield results as anticipated. Client adoption has to be supported by resources that are continuous and orchestrated in an on-going manner. Each step forward in the client journey unlocks the ability for the client to reach the ultimate adoption outcome. Continued efforts are needed to keep the client adopted.</p><blockquote><strong>How do you focus on client adoption for your product in your organization? What part of this post did you find most insightful in your product management role? Leave your comments below</strong>.</blockquote><p>I appreciate any feedback on what topics you would like to read. If you enjoyed reading, please give this some claps below. Click on the Follow to stay tuned to future posts.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5c417380e1ac" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Halftime Break — Reflections on a Career Journey in Product]]></title>
            <link>https://medium.com/swlh/halftime-break-reflections-on-a-career-journey-in-product-2117873f3d52?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/2117873f3d52</guid>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[career-advice]]></category>
            <category><![CDATA[life-lessons]]></category>
            <category><![CDATA[productivity]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Tue, 07 Dec 2021 06:00:32 GMT</pubDate>
            <atom:updated>2021-12-14T12:48:07.721Z</atom:updated>
            <content:encoded><![CDATA[<h3>Halftime Break — Reflections on a Career Journey in Product</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*FNEo18jBpZJYuY02" /><figcaption>Photo by <a href="https://unsplash.com/@craft_ear?utm_source=medium&amp;utm_medium=referral">Jan Tinneberg</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3>Announcing a New Chapter</h3><p>I’ve become a Closer. That is, being a member of <a href="https://www.closedloop.ai/">ClosedLoop.ai</a>. I recently joined the product team at the Austin-based healthcare data science startup. I’m eager and delighted on so many levels to focus on data science product management at this organization. I’ll tell you why below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0aRWL-rmbWdqFkFNwsO6CQ.png" /></figure><ol><li>I’ve focused on analytics for a good part of my career since 2008. My original problem was “how do I find the greatest lift to build an advertising company I was managing?” Thank you Professor <a href="https://www.linkedin.com/in/seethu-seetharaman-39a284/">Seethu Seetharaman</a> and <a href="https://business.rice.edu/">Rice MBA</a> for those solid beginnings. That led to my work at IBM analytics with an awesome team, birthed a computational biology startup out of Texas Medical Center’s (TMC Innovation) digital health <a href="https://www.tmc.edu/innovation/">startup accelerator</a>, and spending <a href="https://medium.com/@danielwu3/journey-into-data-science-and-machine-learning-17df03691ad5">my spare evenings learning</a> with Andrew Ng’s machine learning and DeepLearning.ai courses. I learned python and jupyter notebook, and I really enjoyed it.</li><li>This is an organization that aligns with my personal and professional mission — <em>Making an impact on people’s lives</em>. I’m not surprised that CEO Andrew’s story founding ClosedLoop.ai was similar to mine. Dealing with a family member in the hospital sparked his focus in healthcare. For me, a close friend encountered stomach cancer, which sparked pivotal career decisions. It’s a huge, challenging space to be in. (<em>Below: I share more thoughts on life/career decisions, applicable whether or not you’re in product</em>).</li><li>Product management at this stage means focusing on new products and laying the groundwork for organizational growth at scale. It means moving at a required speed and cadence that can be proved out within months. It really means being clear on the value to clients and users. It looks like idea-to-execution with emphasis on results, rather than an alignment-weighted overhead.</li><li>The team and leadership is phenomenal. ClosedLoop.ai <a href="https://www.closedloop.ai/news/closedloop-ai-wins-1-6-million-cms-artificial-intelligence-health-outcomes-challenge">won</a> the CMS (Centers for Medicare &amp; Medicaid Services) Artificial Intelligence Health Outcomes challenge, beating out a number of established, larger organizations. That’s impressive. However, it really comes down to the team and people. It’s all about relationships. One doesn’t join a company. One joins a team. I’ve spoken to many startup founders and executive leaders. It was clear there was something unique working with the founders, Dave and Andrew, and being close to the leadership. It shows the importance of product to the leadership, and such alignment is key to future success.</li><li>Entrepreneur, Take 5. Lastly, I grew up in a family of entrepreneurs, and the entrepreneurial mindset is a persistent one. Family businesses. Freelance. Founder. Founding team member. I’m grateful for all the experiences regardless of the outcome. What’s my point? Having been in those settings, my senses tell me it’s right timing and role for me. The time is now. I’m excited to be on board!</li></ol><h3>Halftime Break</h3><p><em>Thinking about this new chapter, I think it would be helpful to reflect upon the career choices I’ve made. I’m hopeful that others would find it useful in their career and life journey as well. If you find these to be insightful, feel free to give me a clap and follow me at the bottom.</em></p><p>In 2015, I published <a href="https://medium.com/@danielwu3/career-mind-map-circa-pre-2002-789471f3d97a">14 Lessons Learned Navigating Career Choices</a>, which was based on lessons I learned early in my career. It was based on a moment of clarity in the midst of an enduring obstacle that everyone faces — “What do I do with my career?” It was very much a situation of, “hey this isn’t working out the way I thought. What should I do?” This situation lasted for years. The feeling of being stuck. I had this “aha!” moment, and the output was this mind map and a 5 year plan for my career. I kid you not.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iVndXT9SpzD9uQjpgB37GQ.jpeg" /><figcaption>See link to 14 Lessons Learned for full view</figcaption></figure><p>Truthfully, however, you can’t wait for the light bulb moment to guide your life or career. What I learned is you’ve got to have principals and a North Star. It’s been more than 15 years since I’ve had the mind map. Let’s revisit those original lessons and reflect what I learned. I call this my Halftime Break, meaning I’m halfway through my career.</p><blockquote>What does Daniel 2021 think about Daniel 2006? Was I right? Where was I wrong? What would I tell my younger self to do differently?</blockquote><h3>3 Things That Still Ring True</h3><p>These are a few of the things that are still true in a mid-career as it is in early career —</p><ol><li><strong>Remind yourself what you find purposeful and meaningful.</strong> In every opportunity, I measure it on impact on people, impact on company, and impact that comes from my personal sense of mission and meaning. I’ve used it in my product career in the last four pivotal decisions. By far, be true to yourself. Have that as your North Star (<em>if you don’t have one, see item 1 in the More list</em>).</li><li><strong>Consider your key skills and where else it can be applied. </strong>Focus on the skills you are naturally gifted in and those have interest and can develop further. Look for the roles that would give you satisfaction, not so much for the title.</li><li><strong>Celebrate your decisions. </strong>Make every decision in faith and confidence. Once you do, I’ve to remind myself to live with it. Reward yourself for the decisions no matter how small or significant because it “trains” the brain to avoid indecision (<em>Take this with item 1 in the More list</em>).</li></ol><h3>3 Things to Do More</h3><p>These are the things I would tell my younger self to do more of —</p><ol><li><strong>Take more risks and experiment more early in career. </strong>The saying from Voltaire “perfect is the enemy of good” is so true. The beginning of a career is to learn about oneself and the opportunities out there. Learning is invaluable. I had passed up so many opportunities with potential. Don’t wait for a perfect opportunity. Don’t say no if an opportunity drops in your lap. Try it. If it doesn’t work out, try something else. Nothing wrong with that.</li><li><strong>Ask and reach out to build connections. </strong>During the biodesign fellowship, we would reach out to subject matter experts to learn about their expertise and ask questions. Many times, it was a cold call outreach or LinkedIn message. It amazed me that so many people took our meetings. I’m talking about CEOs down to global health experts. The lesson is be humble. Other people are willing to give their time and expertise to help more than you think. At some point in the future, pay it forward.</li><li><strong>Communication skills over hard skills. </strong>Everyone has different giftedness. Somethings come more naturally. I have a disposition to focus on building things and ease of learning technical, creative, or analytical things. As I advanced in my career, I learned it’s the soft skills that make the bigger difference over the course. Investing in yourself in communicating, writing, speaking, and reading becomes more critical.</li></ol><h3>3 Things to Avoid</h3><ol><li><strong>Following the crowd</strong>. Just because you’ve been taught an idea or see that everyone else seems to be following a particular norm, it doesn’t mean you should stay the course. In my early career, I believed that my career should look a certain way — study well and get a job in a brand name company. When that didn’t happen, I was unhappy for a while. It was stupid to let situations dictate my happiness.</li><li><strong>Demanding yourself unrealistically</strong>. Sometimes it’s not an external circumstance that causes issues. It’s yourself. Letting go of certain expectations is freeing. It liberates your mind from being anchored.</li><li><strong>Neglecting bonds and feeding your soul</strong>. For those prone to driving hard, balance is good. It’s a necessity. I know it’s hard to change. To me, that means spending time with loved ones. Carving out time and energy for others. Learning and applying myself in activities that nourishes the body, mind, and soul.</li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2117873f3d52" width="1" height="1" alt=""><hr><p><a href="https://medium.com/swlh/halftime-break-reflections-on-a-career-journey-in-product-2117873f3d52">Halftime Break — Reflections on a Career Journey in Product</a> was originally published in <a href="https://medium.com/swlh">The Startup</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Thank you and good bye, athenahealth]]></title>
            <link>https://medium.com/@danielwu3/thank-you-and-good-bye-athenahealth-46787a804a3?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/46787a804a3</guid>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Sat, 27 Nov 2021 00:01:12 GMT</pubDate>
            <atom:updated>2021-11-27T00:01:12.640Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6ivzzgnN0miUYai58L4x1A.jpeg" /><figcaption>2018 new employee onboarding at Point Lookout in Maine</figcaption></figure><p>Three years ago I started my journey at athenahealth, a Watertown, MA based health tech company. It was a memorable experience. It started with a three legged trip to Point Lookout, Maine for new employee onboarding. I’ve never had the opportunity to do a remote onboarding in such a location, and my cohort would be one of the last. It was unique and the location was as beautiful as my favorites in the Pacific Northwest.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8gGmGO3vnKXHSIVApl9clQ.jpeg" /><figcaption>The Seaholm building was formerly an Austin Power Plant</figcaption></figure><p>I was based in the Austin, TX Seaholm R&amp;D office, and I initially started in product management in population health. It wasn’t until 6 months later when I would join the scheduling zone where I found a home. I led the Patient Self-Scheduling offering.</p><p>Over the course of the last two to three years, I took an early beta product through to GA and go to market. From 20 or so pilot clients to 3,000 clients who rely on athenahealth’s Patient Self-Scheduling to expand patient access to care — where ever they may be. We went from tens of thousands of appointments booked to more than 1 million this year.</p><p>This achievement is founded upon the dedicated team members who collaborated and partnered to get the product to where it is today. We’ve launched a product for various marketing channels. We’ve integrated in the Patient Portal. We’ve enabled athenahealth to offer telehealth capabilities across athena’s scheduling and in Patient Self-Scheduling. We’ve scaled while responding to client needs during the COVID pandemic.</p><p>I thank the leadership, and my colleagues in product, engineering, design, marketing, analytics, and operations for their support and partnership. I believe a solid groundwork has been laid for future success for Patient Self-Scheduling at athenahealth. I take sincere pride in that and anticipate an outcome that exceeds one with which we laid.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-YN6ysa4IejUiCghpZPsqQ.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=46787a804a3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[700 grams of wheat flour and water]]></title>
            <link>https://medium.com/series/700-grams-of-wheat-flour-and-water-d538edd44587?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/d538edd44587</guid>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Mon, 04 May 2020 04:06:00 GMT</pubDate>
            <atom:updated>2020-05-05T19:32:59.987Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>36 hours later: dispose half and add 1 cup white flour and 1/2 cup water. Mix.</blockquote><p>6 hours later, already growing</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d538edd44587" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Python Machine Learning]]></title>
            <link>https://medium.com/@danielwu3/after-10-months-of-jupyter-notebook-by-notebook-149-lesson-by-lesson-inchwise-progress-i-95f0070e66e9?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/95f0070e66e9</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Fri, 14 Jun 2019 13:35:35 GMT</pubDate>
            <atom:updated>2019-06-14T18:44:54.418Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*YyL0-zrTKcNyDqMGB8MMzw@2x.jpeg" /></figure><p>Python Machine Learning</p><p>After 10 months of Jupyter notebook by notebook, 149 lesson by lesson inchwise progress, I completed the Udemy Python for Data Science and Machine Learning course by Jose Portilla. It was a great follow on to Andrew Ng’s Intro to Machine Learning course that I finished January 2018. In the middle of the course at the beginning of this year, I took a break to analyze the CDC population health dataset doing exploratory data analysis (EDA). Jose’s course was well put together and give it solid recommendations for those looking for broad python based learnings.</p><p>More good things to come soon …</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=95f0070e66e9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Detailed findings of 4 groups of major chronic disease indicators]]></title>
            <link>https://medium.com/data-science/detailed-findings-of-4-groups-of-major-chronic-disease-indicators-53fdeb487155?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/53fdeb487155</guid>
            <category><![CDATA[data-visualization]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[public-health]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[in-depth-analysis]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Mon, 06 May 2019 00:39:01 GMT</pubDate>
            <atom:updated>2019-05-07T22:08:37.191Z</atom:updated>
            <content:encoded><![CDATA[<h3>Four major chronic diseases and premature mortality</h3><h4>Detailed analysis of population health indicators, social determinants of health, and effects on gender and race</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Hmhg7vb4Lye-82vn" /><figcaption>Photo by <a href="https://unsplash.com/@cmdrshane?utm_source=medium&amp;utm_medium=referral">CMDR Shane</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>In <a href="https://towardsdatascience.com/relationships-validated-between-population-health-chronic-indicators-b69e7a37369a">part 2</a> of our CDC Chronic Disease indicator dataset, our analysis revealed several areas with highly correlated interrelationships — indicators within the cardiovascular disease, chronic kidney disease, diabetes, and select indicators in the overarching conditions “social determinants” category. While there are also highly correlated relationships in other areas such as cancer and COPD, we’ll be focusing primarily on the former set in this final blog post.</p><p>While Figure 1 from the previous post looked at the relationships among all the indicators, Table 5 from the previous post showed there were a number of top correlation pairs. By looking at the recurring patterns of indicators by specific topic, we can narrow down the scope of the topics of interest. The same code was used previously was used to build the correlation heatmap in the below six figures (red — higher correlation):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/758/1*mIjNjMn3014NR2oTn0v5Ug.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/758/1*3NMrKvOLJD4yABjxUfvLFw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/758/1*-3CmzsPIwhb_FiTeXeSqAg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/775/1*LcnLiLfY01yV_zp79YB9fw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/775/1*hBQkw5O0dEGz4qsmOC6wjw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/759/1*CCt6gV-2xLeU0JCI5ifA0g.png" /><figcaption>Correlation Heatmap by Topic Pairs (from top left to bottom right): 1) Chronic Kidney (CKD) and Cardiovascular (CVD), 2) CKD and Diabetes (DIA), 3) CKD and Overarching Conditions (OVC), 4) CVD and DIA, 5) CVD and OVC, and 6) DIA and OVC. Red/pink indicates higher positive correlation while green indicates higher negative correlation. See Figures 1 to 6 below for in depth views and descriptions.</figcaption></figure><p>Below we go through each of the correlation heatmaps pair by pair. At the end of section 6, there will be a reference to the actual Questions and QuestionID (QIDs). Feel free to skip if that helps understand the top specific indicator questions.</p><p><strong>Summary of Detailed Findings:</strong></p><p>As we dive into each highly correlated population health chronic indicators by topic and now by individual Question ID pairs, representing the individual indicators, the theme of interrelated chronic illnesses are reinforced. The pattern of comorbidities among diabetes, cardiovascular disease, and chronic kidney disease shows up in general indicators and some more specific indicators. For example, general indicators are broad indicators that appear to capture overall conditions such as DIA1_2 “Mortality due to diabetes reported as any listed cause of death.” These high level indicators with high positive correlation are bound to more discrete mortality outcomes as well, which while obvious, poses questionable ability to bring actionable next steps for individuals. Rather, it implies that the practical route presses towards prevention as a key approach in population health.</p><p>Interestingly, various overarching conditions (OVC) indicators appeared over and over such as OVC5_0 “Premature mortality among adults aged 45–64 years.” We see OVC5_0 paired with all of the top 5 cardiovascular disease indicators, top 3 chronic kidney disease indicators, and the top 4 diabetes indicators. Such relationship tends to speak to the comorbidities of these three interrelated issues and premature mortality for adults 45–64 years of age.</p><p>See below for the findings of the pairs, or jump two sections below as we shift gears to an analysis on the stratifications by gender and race, and the corresponding visualizations.</p><p><strong>1. Chronic Kidney Disease and Cardiovascular Disease (CKD x CVD)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/758/1*mIjNjMn3014NR2oTn0v5Ug.png" /><figcaption>Figure 1. Correlation Heatmap for Chronic Kidney Disease and Cardiovascular Disease</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/341/1*1D8QZHty0agED1aLs9oLgw.jpeg" /><figcaption>Table 1. Top Correlation of Chronic Kidney Disease and Cardiovascular Disease, descending</figcaption></figure><pre>top_corr_pair[(top_corr_pair[&#39;Topic1&#39;] == (&#39;CKD&#39;)) &amp; (top_corr_pair[&#39;Topic2&#39;] == (&#39;CVD&#39;))].head()</pre><p>Chronic kidney disease indicator CKD1_0 is for “Mortality with end-stage renal disease.” On the cardiovascular disease side, CVD1_1 to CVD1_5 are “Mortality from total cardiovascular disease,” “Mortality from diseases of the heart,” “Mortality from coronary heart disease,” “Mortality from heart failure,” and “Mortality from cerebrovascular disease (stroke),” respectively.</p><p><strong>2. Chronic Kidney Disease and Diabetes (CKD x DIA)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/758/1*3NMrKvOLJD4yABjxUfvLFw.png" /><figcaption>Figure 2. Correlation Heatmap for Chronic Kidney Disease and Diabetes</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/346/1*izjffn6X6OuP9e4-gJGZXw.jpeg" /><figcaption>Table 2. Top Correlation of Chronic Kidney Disease and Diabetes, descending</figcaption></figure><p>Chronic kidney disease indicator CKD2_1 and CKD2_2 are “Incidence of treated end-stage renal disease” and “Incidence of treated end-stage renal disease attributed to diabetes,” respectively. On the diabetes side, DIA1_1 and DIA1_2 are “Mortality due to diabetes reported as any listed cause of death” and “Mortality with diabetes ketoacidosis reported as any listed cause of death,” respectively. DIA9_0 is “Hospitalization with diabetes as a listed diagnosis.” DIA4_0 is “Amputation of a lower extremity attributable to diabetes.”</p><p><strong>3. Chronic Kidney Disease and Overarching Conditions (CKD x OVC)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/758/1*-3CmzsPIwhb_FiTeXeSqAg.png" /><figcaption>Figure 3. Correlation Heatmap for Chronic Kidney Disease and Overarching Conditions</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/346/1*gzDND0xOuRR4twWG1uYC5Q.jpeg" /><figcaption>Table 3. Top Correlation of Chronic Kidney Disease and Overarching Conditions, descending</figcaption></figure><p>Overarching Condition indicators OVC5_0 and OVC6_1 are “Premature mortality among adults aged 45–64 years” and “Fair or poor self-rated health status among adults aged ≥ 18 years,” respectively. These indicators are highly correlated with the same Diabetes indicators DIA1_1, DIA1_2, DIA4_0, and DIA9_0. Additionally we’re seeing relatively high correlation between DIA2_1 “High school completion among adults aged 18–24 years” and OVC6_1 with those with fair or poor self-rated health.</p><p><strong>4. Cardiovascular Disease and Diabetes (CVD x DIA)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/775/1*LcnLiLfY01yV_zp79YB9fw.png" /><figcaption>Figure 4. Correlation Heatmap for Cardiovascular Disease and Diabetes</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/339/1*2wcs6oXhNO7xrHdtrwaP7g.jpeg" /><figcaption>Table 4. Top Correlation of Cardiovascular Disease and Diabetes, descending</figcaption></figure><p>When comparing Cardiovascular and Diabetes indicators, the top correlated indicators start with DIA1_1, which is a general QuestionID of any mortality from diabetes, and also include DIA9_0 hospitalization with diabetes in the diagnosis. The related cardiovascular indicators include CVD1_4 (heart failure), CVD3_1 (hospitalization), CVD1_5 (mortality from stroke), CVD1_1 (general mortality from cardiovascular), and CVD1_2 (mortality from diseases of the heart).</p><p><strong>5. Cardiovascular Disease and Overarching Conditions (CVD x OVC)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/775/1*hBQkw5O0dEGz4qsmOC6wjw.png" /><figcaption>Figure 5. Correlation Heatmap for Cardiovascular Disease and Overarching Conditions</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/328/1*-BYvuRF5t6XRo1Ksozm0aQ.jpeg" /><figcaption>Table 5. Top Correlation of Cardiovascular Disease and Overarching Conditions, descending</figcaption></figure><p>When comparing highly correlated Cardiovascular Disease and Overarching Conditions indicator OVC5_0 “premature mortality 45–64 years adults,” the top correlating indicators are belong to the group of CVD1_X Question IDs (QID1), ranging from heart disease to stroke.</p><p><strong>6. Diabetes and Overarching Conditions (DIA x OVC)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/759/1*CCt6gV-2xLeU0JCI5ifA0g.png" /><figcaption>Figure 6. Correlation Heatmap for Diabetes and Overarching Conditions</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/340/1*qJrY2ToqhxK6UUObNoet4Q.jpeg" /><figcaption>Table 6. Top Correlation of Diabetes and Overarching Conditions, descending</figcaption></figure><p>Overarching Condition indicators OVC5_0 and OVC6_1 are “Premature mortality among adults aged 45–64 years” and “Fair or poor self-rated health status among adults aged ≥ 18 years,” respectively. These indicators are highly correlated with the same Diabetes indicators DIA1_1, DIA1_2, DIA4_0, and DIA9_0. Additionally we’re seeing relatively high correlation between DIA2_1 “High school completion among adults aged 18–24 years” and OVC6_1 with those with fair or poor self-rated health.</p><p><strong>Reference for QuestionID Indicator Codes</strong></p><p>Since Figures/Tables 1 to 6 visualized or listed the relationships, it would be helpful to take a look again at the respective questions for the Question IDs (QID1/QID2) to better understand the specific indicators. Using groupby() method on the df_new dataframe and then .loc[] to pull out the respective topics, I’m able to view a list of the QuestionID and Questions.</p><pre>df_new_QTQLY = df_new[[&#39;QuestionID&#39;,&#39;Topic&#39;,&#39;Question&#39;,&#39;LocationAbbr&#39;,&#39;YearStart&#39;]].groupby([&#39;Topic&#39;,&#39;QuestionID&#39;,&#39;Question&#39;]).count()</pre><pre>df_new_QTQLY.loc[&#39;Cardiovascular Disease&#39;]</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/491/1*sxKcaGecNOzVka-lYvUxJA.jpeg" /><figcaption>Table 7a. QuestionID for Chronic Kidney Disease Indicators</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/801/1*OHkYY3nglwVnxef6ZWIlmw.jpeg" /><figcaption>Table 7b. QuestionID for Cardiovascular Disease Indicators</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/478/1*haBtj93XtdCIZcbnHpL4ug.jpeg" /><figcaption>Table 7c. QuestionID for Overarching Condition Indicators</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/685/1*PFcn4GFyZx6HU-R_scSXDg.jpeg" /><figcaption>Table 7d. QuestionID for Diabetes Indicators</figcaption></figure><p><strong>Stratification Analysis on Adult Premature Mortality — Gender</strong></p><p>Since OVC5_0 indicator “Premature mortality of adults 45–64 years of age” appear repeatedly, let’s look into what this indicator can tell us by stratification of gender and race. We’re going to create df_new_OVC5_0_gender dataframe based on df_new based on the following parameters:</p><pre>df_new_OVC5_0_gender = df_new[ (df_new[&#39;QuestionID&#39;] == &#39;OVC5_0&#39;) &amp; (df_new[&#39;StratificationCategory1&#39;] == &#39;Gender&#39;) &amp; <br>(df_new[&#39;DataValueUnit&#39;] == &#39;cases per 100,000&#39;)]</pre><pre>df_new_OVC5_0_gender.info()</pre><p>In the above dataframe, df_new[‘QuestionID’] == ‘OVC5_0’ takes the subset of data related to our specific indicator. However, this includes stratifications for both race and gender. To look at the gender stratification, df_new[‘StratificationCategory1’] == ‘Gender’ is included. The remaining dataframe shows in DataValueType to include either a ‘Number’ or an ‘Age adjusted rate.’ The ‘Number’ represents a crude rate, which is the “total burden of a health outcome to a community.” The ‘age adjusted rate’ is used to make more fair comparisons among groups whether it’s age, gender, or race. (More detailed explain <a href="https://health.mo.gov/data/mica/CDP_MICA/AARate.html">here</a>). It appears that when DataValueType = ‘Age adjusted rate,’ DataValueUnit = ‘cases per 100,000’ so I’ve included df_new[‘DataValueUnit’] == ‘cases per 100,000.’ The count shows that the OVC topic and StratificationCategory1 is not as populated as the full dataset which is expected:</p><pre>&lt;class &#39;pandas.core.frame.DataFrame&#39;&gt;<br>Int64Index: 1020 entries, 268047 to 403039<br>Data columns (total 16 columns):<br>YearStart                    1020 non-null int64<br>LocationAbbr                 1020 non-null object<br>Topic                        1020 non-null object<br>Question                     1020 non-null object<br>DataValueUnit                1020 non-null object<br>DataValueType                1020 non-null object<br>DataValueAlt                 1020 non-null float64<br>StratificationCategory1      1020 non-null object<br>Stratification1              1020 non-null object<br>LocationID                   1020 non-null int64<br>TopicID                      1020 non-null object<br>QuestionID                   1020 non-null object<br>DataValueTypeID              1020 non-null object<br>StratificationCategoryID1    1020 non-null object<br>StratificationID1            1020 non-null object<br>QuestionAbbr                 1020 non-null object<br>dtypes: float64(1), int64(2), object(13)</pre><pre>df_new_OVC5_0_gender.head()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1010/1*sK1QkiASpm06DvDOXD0FSA.jpeg" /><figcaption>Table 8a. Stratification by Gender Dataframe</figcaption></figure><p>Using the groupby summary, we see that the data available runs for 5 years as opposed to the entire 15 years of the overall dataset. The following will also plot out the trends and in a seaborn barplot.</p><pre>df_new_OVC5_0_gender1 = df_new_OVC5_0_gender.groupby([&#39;Stratification1&#39;,&#39;YearStart&#39;])<br>df_new_OVC5_0_gender1.mean().drop(&#39;LocationID&#39;,axis=1).round()</pre><pre>plt.figure(figsize=(16, 6))<br>sns.barplot(x=&#39;YearStart&#39;,y=&#39;DataValueAlt&#39;,data=df_new_OVC5_0_gender,hue=&#39;Stratification1&#39;,ci=None,saturation=0.7)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/259/1*InE2N5DddR_oB9SYp8-N-g.jpeg" /><figcaption>Table 8b. Groupby on the Stratification Gender dataset</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/947/1*oM-3PCC_ADGerSq96PjEnQ.png" /><figcaption>Figure 8. 2010–2014 Trends on Premature Mortality by Gender, per 100K cases</figcaption></figure><p>Over the course of 2010 to 2014, the data shows that Premature Mortality of 45–64 years of age population to slightly increase year over year. Female population premature mortality increases 5.4% and male population at 3.4%. The vertical axis corresponds to DataValueAlt which is in units of 100K cases. For example, there are 485K cases in 2014 for Female (38%) while there are 784K cases in the same year for Male (62%). Surprisingly, I anticipated the premature mortality rate to increase faster for Male over Female overall.</p><p>While the correlation pairs gave an overall look, let’s compare indicators between Male and Female populations in the same fashion as by correlation pairs. In the original df_new[‘QuestionID’], there were 201 unique indicators across the dataset. Since we’re focused on the age-adjusted group, our df_new_gender2[‘QuestionID’] shows there are 13 unique indicators. Keeping that in mind, the following analysis shows a subset of indicators. We’re going to start with modifying the dataframe df_new_gender for the following gender and a potential location analysis, thereby dropping several columns that are not useful for this.</p><pre>df_new_gender2 = df_new_gender.drop(columns=[&#39;Question&#39;,&#39;Topic&#39;,&#39;DataValueUnit&#39;,&#39;DataValueType&#39;,&#39;StratificationCategory1&#39;,&#39;TopicID&#39;,&#39;DataValueTypeID&#39;,&#39;StratificationCategoryID1&#39;,&#39;StratificationID1&#39;,&#39;QuestionAbbr&#39;])<br>df_new_gender2.head()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/1*USu6kxkQIMCqR1SPw-PFmA.png" /><figcaption>Table 9a. Working Dataframe with Stratification1 (Gender) and Location</figcaption></figure><p>In order to create a table to visualize the correlation pairs by gender, let’s take Table 9a and apply the pivot_table method. Let’s create a dataframe called df_new_gender2_loc_qid where the rows are for each location and columns are each indicator. Since we want to see the data by gender (Stratification1), there needs to be a column with a label Male or Female for each location. Also, since we’re looking at each row by location, it makes sense to summarize the DataValueAlt values for each indicator as an average so we’ll pass in the parameter aggfunc=np.mean into the pivot_table.</p><p>As this pivot table is created, I encountered issues with LocationIDs or QuestionIDs that don’t have data, resulting in NA values. When this happens, we can assess whether to keep only rows/columns that have values by dropping all the rows and/or columns with NA values. As I did this, the resulting dataframe was severely reduced. Instead, by using the .fillna() with a mean using df_new_gender_loc_qid.mean(), the NA values are replaced with an average. Table 9b is the top 5 rows of the adjusted pivot table.</p><pre>df_new_gender2_loc_qid = df_new_gender2.pivot_table(values=&#39;DataValueAlt&#39;,index=[&#39;LocationID&#39;,&#39;Stratification1&#39;],columns=[&#39;QuestionID&#39;],aggfunc=np.mean)<br># Create pivot table with rows by Location, columns of each QuestionID, values - DataValueAlt mean by Location</pre><pre>df_new_gender2_loc_qid.reset_index(level=&#39;Stratification1&#39;,inplace=True)<br>df_new_gender2_loc_qid.fillna(df_new_gender2_loc_qid.mean(),inplace=True)<br>df_new_gender2_loc_qid.head()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/944/1*eyJRJQ9p5vNeknyNctQ_7w.png" /><figcaption>Table 9b. Pivot Table by Location and Gender Across 13 Indicators (QuestionIDs)</figcaption></figure><p>With seaborn pairplot on df_new_gender2_loc_qid, the hue parameter allows us to render the plot by gender. Similar to <a href="https://cdn-images-1.medium.com/max/2400/1*Ip5aiJBTRzCrnvGjp-NtCQ.png">Figure 1</a> from the <a href="https://towardsdatascience.com/relationships-validated-between-population-health-chronic-indicators-b69e7a37369a">Part 2 blog</a> post, we have a plot of each of the 13 indicators along the same X and Y axes. Instead of correlation, we have the average values from DataValueAlt by gender and state (LocationID) for these 13 indicators: ALC6_0 (Alcohol: Chronic Liver Disease Mortality), CKD1_0 (Chronic Kidney Disease: End Stage Renal Disease Mortality), 2 Chronic Obstructive Pulmonary Disease (COPD), 5 Cardiovascular Disease (CVD), 2 Diabetes (DIA), OLD1_0, and OVC5_0 (Overarching Conditions: Premature Mortality Aged 45–64 Years). OVC5_0 came up in the prior sections as a highly correlated indicator of interest.</p><pre>sns.pairplot(df_new_gender2_loc_qid,hue=&#39;Stratification1&#39;,palette=&#39;RdBu_r&#39;)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zQgbtisDVZCjdsTmduhLdA.png" /><figcaption>Figure 9. Pairplot Across 13 Indicators Comparing Gender (Blue — Female, Orange — Male)</figcaption></figure><p>As mentioned previously about Figure 9 above, OVC5_0 is the Premature Mortality Aged 45–64 Years indicator of interest. Along the bottom row of Figure 9, the pairplot generally shows a distinct distribution between Male (orange) and Female (blue) populations. In nearly all cases when we compare each plot along the bottom row (OVC5_0 to the other indicators), we see Male populations with higher average DataValueAlt values for most locations, which reflects the generally, consistently poorer population health conditions across states for Male.</p><p><strong>Stratification Analysis on Adult Premature Mortality —Race</strong></p><p>Similar to stratification by gender, the analysis on stratification by race will use the following conditions to filter the data. We’ll create a new dataframe df_new_OVC5_0_race based on OVC5_0, aged based prevalence using “cases per 100,000,” and the StratificationCategory1 of “Race/Ethnicity.”</p><pre>df_new_OVC5_0_race = df_new[ (df_new[&#39;QuestionID&#39;] == &#39;OVC5_0&#39;) &amp; (df_new[&#39;DataValueUnit&#39;] == &#39;cases per 100,000&#39;) &amp;(df_new[&#39;StratificationCategory1&#39;] == &#39;Race/Ethnicity&#39;)]</pre><pre>df_new_OVC5_0_race1 = df_new_OVC5_0_race.groupby([&#39;Stratification1&#39;,&#39;YearStart&#39;]).mean().round(0)<br>df_new_OVC5_0_race1</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/334/1*sRQgHneMN6Bk104hGZlMtA.jpeg" /><figcaption>Table 10. Stratification by Race with Average DataValueAlt Values</figcaption></figure><p>Further we’ll use the .groupby() method to summarize the subset dataframe df_new_OVC5_0_race into a table based on the race, year, and the DataValueAlt. The categories for race in Stratification1 gives us 5 categories: American Indian or Alaska Native; Asian or Pacific Islander; Black, non-Hispanic; Hispanic; and White, non-Hispanic. For YearStart, we have 5 years of data since this was the result of the available stratification data and the use of the aged-based prevalence. DataValueAlt is an average value of the locations, rounded off to the whole digits. For example,there are 348 cases of Premature Mortality in 45–64 years aged persons per 100K for the Hispanic population in 2010.</p><p>To visualize Table 10, seaborn barplot will display each of the race stratifications and the values of DataValueAlt. The parameters for the barplot include passing the x=YearStart, y=DataValueAlt, and hue=Stratification1 from data=df_new_OVC5_0_race. With plt.legend(), we’re able to fine tweak where the legend gets rendered, and I prefer outside plot which uses the bbox_to_anchor parameter.</p><pre>plt.figure(figsize=(16, 7))<br>sns.barplot(x=&#39;YearStart&#39;,y=&#39;DataValueAlt&#39;,data=df_new_OVC5_0_race,hue=&#39;Stratification1&#39;,ci=None,saturation=0.7)<br>plt.legend(loc=&#39;best&#39;,bbox_to_anchor=(0.5, 0,.73, .73))</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sCJDJ_qMbhD3oq7SsRxGcg.png" /><figcaption>Figure 10. 2010–2014 Premature Mortality Trends Across Race Stratification</figcaption></figure><p>From both Table 10 and Figure 10, the data shows that premature mortality is slightly increasing year over year with incremental increases. Premature mortality of adults aged 45–64 years of age impact the “Black, non-Hispanic” and the “American Indian or Alaska Native” populations most severely. On the other spectrum, we see “Asian or Pacific Islander” and “Hispanic” populations to be least severe. By comparison, we’re seeing the premature mortality numbers for “Black, non-Hispanic” population to be more than double of the “Hispanic” population.</p><p>While it’s possible to hypothesize the correlations with the premature mortality numbers for each stratification, it’s beyond the scope of this dataset to tie the descriptive with the causation. Figure 10 gives a general understanding of the seriousness of the factors impacting the overall population, which can be complex and multi-faceted.</p><p>Another way of looking at the race stratification is by a stacked bar chart. Again, a pivot_table() with the same data from df_new_OVC5_0_race is a starting point for the bar chart, shown in Table 11.</p><pre>df_new_OVC5_0_race2 = df_new_OVC5_0_race.pivot_table(values=&#39;DataValueAlt&#39;,index=&#39;YearStart&#39;,columns=&#39;Stratification1&#39;).round()<br>df_new_OVC5_0_race2 </pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/788/1*H_GKJMfVJfZRC3Z8SilsoQ.jpeg" /><figcaption>Table 11. Stratification by Race and Year</figcaption></figure><p>Based on Table 11, we want to create columns of cumulative values instead of the specific values for each stratification. I’m going to modify df_new_OVC5_0_race2 by updating the values for subsequent columns as the sum of the previous columns. For example the values in the column for Asian or Pacific Islander column will be the sum of American Indian or Alaska Native and the Asian or Pacific Islander. Repeat this for the next three columns. With a few additional tweaks of dropping columns to keep the new cumulative columns and resetting the index, the new dataframe df_new_OVC5_0_race2c can be used in the following barplot.</p><pre>sns.set_style(&#39;darkgrid&#39;)<br>plt.figure(figsize=(8, 8))<br>sns.barplot(x=&#39;YearStart&#39;, y=&#39;cum_White&#39;, data=df_new_OVC5_0_race2c, color=&#39;blue&#39;,ci=None,saturation=0.7)<br>sns.barplot(x=&#39;YearStart&#39;, y=&#39;cum_Hispanic&#39;, data=df_new_OVC5_0_race2c, color=&#39;purple&#39;,ci=None,saturation=0.7)<br>sns.barplot(x=&#39;YearStart&#39;, y=&#39;cum_Black&#39;, data=df_new_OVC5_0_race2c, color=&#39;red&#39;,ci=None,saturation=0.7)<br>sns.barplot(x=&#39;YearStart&#39;, y=&#39;cum_Asian&#39;, data=df_new_OVC5_0_race2c, color=&#39;green&#39;,ci=None,saturation=0.7)<br>sns.barplot(x=&#39;YearStart&#39;, y=&#39;cum_AmIndianAK&#39;, data=df_new_OVC5_0_race2c, color=&#39;orange&#39;,ci=None,saturation=0.7)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/506/1*hEfHgb7RwvRVEBJQ1TW-7w.png" /><figcaption>Figure 11. Stacked Bar Chart of Stratification by Race and Year</figcaption></figure><p>Since the year to year changes for each race stratification are not significantly different, the bars are mostly similar. However, this bar chart gives another view into the overall premature mortality growth from 2010 to 2014. The chart also provides a view of the proportion of the premature mortality amongst the categories of race. The colors of the hues use the same legend as in Figure 10: Blue (White, non-Hispanic), Purple (Hispanic), Red (Black non-Hispanic), Green (Asian Pacific Islander), and Yellow/Orange (American Indian or Alaska Native).</p><p>I hope you’ve enjoyed following the reading on the US Chronic Indicators population dataset from the CDC. This wraps up our series. Follow me for more digital health data science related topics. Feel free to suggest any ideas or recommendations!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=53fdeb487155" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/detailed-findings-of-4-groups-of-major-chronic-disease-indicators-53fdeb487155">Detailed findings of 4 groups of major chronic disease indicators</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Relationships validated between population health chronic indicators]]></title>
            <link>https://medium.com/data-science/relationships-validated-between-population-health-chronic-indicators-b69e7a37369a?source=rss-a2bc7fe3d2d4------2</link>
            <guid isPermaLink="false">https://medium.com/p/b69e7a37369a</guid>
            <category><![CDATA[digital-health]]></category>
            <category><![CDATA[public-health]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[in-depth-analysis]]></category>
            <category><![CDATA[python]]></category>
            <dc:creator><![CDATA[Daniel Wu]]></dc:creator>
            <pubDate>Thu, 21 Mar 2019 14:16:10 GMT</pubDate>
            <atom:updated>2019-05-13T01:54:06.570Z</atom:updated>
            <content:encoded><![CDATA[<h4>The ties among diabetes, chronic kidney disease, and cardiovascular disease</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*HWK8EGnYG0ZZHzJN" /><figcaption>Photo by <a href="https://unsplash.com/@v2osk?utm_source=medium&amp;utm_medium=referral">v2osk</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>In the <a href="https://medium.com/@danielwu3/inspecting-a-cdc-chronic-disease-dataset-e1685a6b525a">last story</a>, we started looking into a 15 year chronic disease dataset from the U.S. Center for Disease Control and Prevention, or CDC. The beginnings of the exploratory data analysis started with understanding the columns and rows of data and what was relevant for further analysis.</p><p>In this post, we are going to dig deeper to understand these 400K rows and 17 categories of topics, which requires a bit of data wrangling of the dataframe into a format for pivot table summary and visualization. After looking at the previous df_new.head(), I don’t think we will use the following columns, which results in a smaller set of 15 columns:</p><pre>df_new = df_new.drop(columns=[&#39;YearEnd&#39;,&#39;LocationDesc&#39;,&#39;DataSource&#39;,&#39;DataValue&#39;,&#39;DataValueFootnoteSymbol&#39;,&#39;DatavalueFootnote&#39;,&#39;LowConfidenceLimit&#39;,&#39;HighConfidenceLimit&#39;,&#39;GeoLocation&#39;])</pre><pre>&lt;class &#39;pandas.core.frame.DataFrame&#39;&gt;<br>RangeIndex: 403984 entries, 0 to 403983<br>Data columns (total 15 columns):<br>YearStart                    403984 non-null int64<br>LocationAbbr                 403984 non-null object<br>Topic                        403984 non-null object<br>Question                     403984 non-null object<br>DataValueUnit                374119 non-null object<br>DataValueType                403984 non-null object<br>DataValueAlt                 273666 non-null float64<br>StratificationCategory1      403984 non-null object<br>Stratification1              403984 non-null object<br>LocationID                   403984 non-null int64<br>TopicID                      403984 non-null object<br>QuestionID                   403984 non-null object<br>DataValueTypeID              403984 non-null object<br>StratificationCategoryID1    403984 non-null object<br>StratificationID1            403984 non-null object<br>dtypes: float64(1), int64(2), object(12)</pre><p><strong>Table 1: Summarizing by Topic, Question, and Value</strong></p><p>Using a pivot table (much like Excel), let’s summarize each Question and its related values. To inform each of the questions and values, we’ll create multiple indexes from the category of each question (Topic), the QuestionID, Question, the unit of the data (DataValueUnit), and the type of the data relevant to the stratifications later (DataValueType). Pandas Pivot_table requires numeric values, and defaults to averages (numpy.mean). Further, I’ve dropped columns LocationID and YearStart since I won’t be using these later, and rounded to two decimals for readability:</p><pre>df_QD = df_new.pivot_table(index=[&#39;Topic&#39;,&#39;QuestionID&#39;,&#39;Question&#39;,&#39;DataValueUnit&#39;,&#39;DataValueType&#39;],columns=None,dropna=True)</pre><pre>df_QD.drop(columns=[&#39;LocationID&#39;,&#39;YearStart&#39;]).round(2).head(25)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/996/1*DdWmiqClLNlrmUs966tErw.png" /><figcaption>Table 1. Pivot Table showing the Question and Values</figcaption></figure><p><strong>Table 2: Using Groupby with Stratification, Question, and Value</strong></p><p>Alternative to the pivot_table, I can create a similar summary table using groupby(). I want to show a bit more information on the stratification in the resulting dataframe.</p><pre>df_unittype_table1 = df_new.groupby([&#39;Topic&#39;,&#39;QuestionID&#39;,&#39;Question&#39;,&#39;StratificationID1&#39;,&#39;Stratification1&#39;,&#39;DataValueUnit&#39;,&#39;DataValueType&#39;]).mean().round(2)</pre><pre>df_unittype_table1.drop(columns=[&#39;YearStart&#39;,&#39;LocationID&#39;])</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*FLWkDkulX75WqZKqXQZv3A.png" /><figcaption>Table 2. Using groupby() to summarize stratification</figcaption></figure><p><strong>Table 3: Summarizing by Question and Location</strong></p><p>An interesting approach would showing the data by each state (LocationAbbr) with the data values. The following pivot table includes additional indexes that provide context to the numbers such as the unit and type.</p><pre>df_new_qloc = df_new.pivot_table(values=&#39;DataValueAlt&#39;,index=[&#39;Topic&#39;,&#39;QuestionID&#39;,&#39;Question&#39;,&#39;DataValueUnit&#39;,&#39;DataValueType&#39;], columns=&#39;LocationAbbr&#39;,aggfunc=&#39;mean&#39;,dropna=True).round(2)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*vlP3a0fV7Rl3qPQ1KzeMxw.png" /><figcaption>Table 3. Pivot Table of Question and Location</figcaption></figure><p><strong>Visualizing Correlation Among All Indicators</strong></p><p>Now we’ve seen the tables present the data values by question, by stratification, and by location. That’s still a lot of information! Let’s understand what the question indicators say about itself. What are the relationships between each pair of question indicators? This is where visualization of the data will paint a picture to understand the overall relationships.</p><p>Originally, I started out using the pivot_table from Table 3 but the axis labels were too long and difficult to read. Backing up slightly, we’re going to create a different pivot_table (df_new_qloc2) based on the new the column QuestionAbbr to take the first 37 characters of Question:</p><pre>df_new[&#39;QuestionAbbr&#39;] = df_new[&#39;Question&#39;].str[:37]</pre><pre>df_new_qloc2 = df_new.pivot_table(values=&#39;DataValueAlt&#39;,index=[&#39;Topic&#39;,&#39;QuestionID&#39;,&#39;QuestionAbbr&#39;], columns=&#39;LocationAbbr&#39;,aggfunc=&#39;mean&#39;,dropna=True).round(2)</pre><p>By transposing df_new_qloc2 pivot table, we’re ready to apply the correlation method .corr() and visualize the data. Using the code from the <a href="https://seaborn.pydata.org/examples/many_pairwise_correlations.html">seaborn</a> documentation, we can plot a correlation matrix heatmap. This visualizes each pair of indicators to understand where the positive correlation pairs reside. Since a correlation matrix yields duplicate views above and below the diagonal, we’ll mask the upper half of the heatmap for simplicity by creating an array of zeros using np.zeros_like() and returning the indices of the region above the diagonal using triu_indices_from(). Using matplotlib, we set a large figure size to allow zooming in on the jupyter notebook. The cmap sets the colormap for the figure (more options <a href="https://seaborn.pydata.org/generated/seaborn.diverging_palette.html">here</a>). Lastly, sns.heatmap() renders the heatmap with the mask.</p><pre># Table 4 - Using Table 3 with question indicators as columns<br>new_qloc2_corr = df_new_qloc2.transpose().corr()</pre><pre># Generate a mask for the upper triangle<br>mask = np.zeros_like(new_qloc2_corr, dtype=np.bool)<br>mask[np.triu_indices_from(mask)] = True</pre><pre># Set up the matplotlib figure<br>f, ax = plt.subplots(figsize=(150, 150))<br>f.suptitle(&#39;All Chronic Indicators Correlation Heatmap&#39;, x=0.4,y=0.85,fontsize=150)<br>ax.tick_params(labelsize=50)</pre><pre># Generate a custom diverging colormap<br>cmap = sns.diverging_palette(190, 10, as_cmap=True)</pre><pre># Draw the heatmap with the mask and correct aspect ratio<br>sns.heatmap(new_qloc2_corr, mask=mask, cmap=cmap, vmax=.3, center=0,square=True, linewidths=2, cbar_kws={&quot;shrink&quot;: .3})</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ip5aiJBTRzCrnvGjp-NtCQ.png" /><figcaption>Figure 1. All Indicators Correlation Heatmap (Higher Correlation in Pink)</figcaption></figure><p>Even with some tweaking, the information is dense and could be more user friendly to read (we’ll address this in the next post). Still, the heatmap axes show the topic, QuestionID, and the abbreviated question QuestionAbbr. We can see some areas with more positive correlation in pink and more negative in green. Visually, there are areas in the Cardiovascular (CVD) and Cancer (CAN) that appear to have higher correlation. It’s helpful to see this presentation but ultimately, we will want to have a more granular look by creating a list of top correlation.</p><p><strong>Table 5. Top Correlation Table of Indicator Pairs</strong></p><p>Taking only the QuestionID, LocationAbbr, and DataValueAlt out of df_new, I created new_qloc2_1_corr table. Using linear algebra, we can vectorize the arrays by multiplying the new_qloc2_1_corr correlation dataframe and the lower half of an array the same shape as the prior dataframe but with values of one’s. The .stack() puts the array into a list and .sort_values() sorts the list into the descending order. Top_corr consists of about 36k items.</p><pre>top_corr = (new_qloc2_1_corr * np.tril(np.ones(new_qloc2_1_corr.shape),-1)).stack().sort_values(by=[&#39;DataValueAlt&#39;],ascending=False)</pre><p>With a few additional steps to reset the indexes, change the column names, and adding a few new columns based on the QuestionID, we have a dataframe with the top correlation pairs. Since there many duplicate pairs by the topic, I’m going to de-duplicate and not include the same pairs of indicators by topic category which results in about 33k items:</p><pre>top_corr_pair_dd = []</pre><pre>for row in range(len(top_corr_pair)):</pre><pre>     if top_corr_pair.loc[row][&#39;Topic1&#39;] != top_corr_pair.loc[row].     <br>     [&#39;Topic2&#39;]:<br>          top_corr_pair_dd.append(top_corr_pair.loc[row])</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/368/1*hPldbsjvV6tpqs5SiHnPFQ.jpeg" /><figcaption>Table 5. Top Correlation by Topic Descending</figcaption></figure><p>At this point, we see themes emerging from the indicators. Table 5 shows us that there are certain patterns based on the topic categories. For example, those QuestionIDs (QID1 and QID2 columns) within Chronic Kidney Disease (CKD) and Cardiovascular Disease (CVD); those within CKD and Diabetes (DIA); those within CVD and DIA; those within Overarching Conditions (OVC) and prior indicators; and finally, those within Chronic Obstructive Pulmonary Disease (COP) with DIA and CVD.</p><p><strong>Does this make sense overall that we are seeing relationships among these indicators such as CVD, CKD, and DIA?</strong></p><p>Cardiovascular disease includes a number of heart related diseases, including arrhythmia (improper irregular heart beating), high blood pressure (high force against artery walls), cardiac arrest (sudden loss of function), coronary artery disease (damage of major blood vessels), and congestive heart failure (chronic condition where heart doesn’t pump well). It’s a very common disease that stems from blocked blood vessels and impacts 1 in 4 Americans. According to the <a href="https://www.cdc.gov/heartdisease/facts.htm">CDC</a>, risk factors that increase this disease include diet, obesity, diabetes, excessive alcohol use and physical inactivity.</p><p>Chronic kidney disease is the lengthy disease of the kidney and its loss of function leading to failure, also known as end-stage renal disease (ESRD). It’s also a common disease that affects 1 in 7 Americans. According to the <a href="https://www.mayoclinic.org/diseases-conditions/chronic-kidney-disease/symptoms-causes/syc-20354521">Mayo Clinic</a>, risk factors that increase this disease include diabetes, high blood pressure, smoking, cardiovascular disease, older age, and certain ethnic backgrounds. In patients with kidney disease, cardiovascular disease tends to be “underdiagnosed and undertreated” (Said S, Hernandez GT. The link between chronic kidney disease and cardiovascular disease. <em>J Nephropathol</em>. 2014;3(3):99–104.). When the kidneys don’t function optimally, this requires the heart and cardiovascular system to work harder according to the <a href="http://www.kidneyfund.org/kidney-disease/chronic-kidney-disease-ckd/complications/heart-disease/#heart-disease-ckd-connection">American Kidney Fund</a>.</p><p>Diabetes mellitus relates to the disregulation of the glucose in the body, and includes a few diseases such as Type 1, Type 2 and gestational diabetes. While Type 1 cause is unknown such as family background and environment, gestational diabetes usually comes and goes with pregnancy. Type 2 diabetes has similar risk factors to above diseases, including physical inactivity, age, family background, high blood pressure, weight, polycystic ovary syndrome, and abnormal cholesterol / triglyceride levels (Source: <a href="https://www.mayoclinic.org/diseases-conditions/diabetes/symptoms-causes/syc-20371444">Mayo Clinic</a>). Having diabetes leads to various complications including cardiovascular and chronic kidney disease.</p><blockquote>The answer is yes! This dataset is consistent with the literature showing that there is high correlation among cardiovascular disease, chronic kidney disease, and diabetes.</blockquote><p>In the <a href="https://medium.com/p/detailed-findings-of-4-groups-of-major-chronic-disease-indicators-53fdeb487155?source=email-a2bc7fe3d2d4--writer.postDistributed&amp;sk=92d3364be11a826ba1ed44a31600b985">final blog post</a> of the series, we will continue this insight to better understand the following:</p><ul><li>What are the Overarching Conditions?</li><li>What are additional highlights on the specific top indicators?</li><li>What trends can we see by year and by stratification?</li></ul><p>Hope you’ve enjoyed reading to this point and feel free to drop me some ideas and suggestions!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b69e7a37369a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/relationships-validated-between-population-health-chronic-indicators-b69e7a37369a">Relationships validated between population health chronic indicators</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>