<?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 Yordanos Goshu on Medium]]></title>
        <description><![CDATA[Stories by Yordanos Goshu on Medium]]></description>
        <link>https://medium.com/@yag3?source=rss-e93f994af386------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*MPqt7GLBmZ0b0QpP</url>
            <title>Stories by Yordanos Goshu on Medium</title>
            <link>https://medium.com/@yag3?source=rss-e93f994af386------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 12:06:24 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@yag3/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[Grammarly for doctors]]></title>
            <link>https://medium.com/@yag3/grammarly-for-doctors-ebd4d37e7a4?source=rss-e93f994af386------2</link>
            <guid isPermaLink="false">https://medium.com/p/ebd4d37e7a4</guid>
            <dc:creator><![CDATA[Yordanos Goshu]]></dc:creator>
            <pubDate>Fri, 20 Mar 2020 02:04:15 GMT</pubDate>
            <atom:updated>2020-03-20T02:04:15.123Z</atom:updated>
            <content:encoded><![CDATA[<p>As we scope the future direction of our project we have to consider how we’re going to build a product that has value through differentiation and usefulness. These are the metrics that we are currently using to determine our next move: Grammarly for doctors.</p><p>We have noticed that the only to get to the source of mistakes in coding is to originally write structured doctor notes in a user-friendly way. That is why we are building an autocomplete platform for doctors to use when taking notes on their patients!</p><p>We’re excited for the outcome!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ebd4d37e7a4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The bottleneck: integrating with EHRs]]></title>
            <link>https://medium.com/@yag3/the-bottleneck-integrating-with-ehrs-602e599eb870?source=rss-e93f994af386------2</link>
            <guid isPermaLink="false">https://medium.com/p/602e599eb870</guid>
            <dc:creator><![CDATA[Yordanos Goshu]]></dc:creator>
            <pubDate>Fri, 06 Mar 2020 22:06:20 GMT</pubDate>
            <atom:updated>2020-03-06T22:06:20.392Z</atom:updated>
            <content:encoded><![CDATA[<p>One issue we have identified is that our product in some shape or form will need to integrate with an electronic health record system. The reason we have made this assumption is because building out a system that integrates physicians note taking platform to the billing system is a hard system to break into because of the sensitivity of patient data. Thus as the head of Growth I have begun to assemble a list of smaller EHR systems that would be suitable to integrate with. The reason that we are aiming for smaller EHR’s is that larger ones monopolize the the space and they do not want to let outside players in.</p><p>Stay tuned to hear about what they say for possible collaborations!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=602e599eb870" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Navigating the medical room]]></title>
            <link>https://medium.com/@yag3/navigating-the-medical-room-adcd85b70b76?source=rss-e93f994af386------2</link>
            <guid isPermaLink="false">https://medium.com/p/adcd85b70b76</guid>
            <dc:creator><![CDATA[Yordanos Goshu]]></dc:creator>
            <pubDate>Fri, 06 Mar 2020 21:23:18 GMT</pubDate>
            <atom:updated>2020-03-06T21:23:18.621Z</atom:updated>
            <content:encoded><![CDATA[<p>A recent interview I had was with an Anesthesiologist. From her perspective the process of having to take down medical billing notes while being in the procedure room presents itself as quite the burden. In certain cases, causing medical billing mistakes and putting the health of the patient at risk as it detracts the physicians from being fully focused on the patient.</p><p>Additionally, depending on the medical billing system available to the physciain it might take them more time to put information into the system. To this end, she said that systems like Epic have done a good job to make this process more user friends. Specifically, she said that Epic sits with the department heads and outline all the possible choices that have to be selected for each type of surgery. By amortizing this process, future physicians only have to go to a drop down menu and select the dosage. Also to this end, they said there are already tools in place to transcribe the entire operation but maybe something that translates the text into structured billing information may be useful.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=adcd85b70b76" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Health insurance cycle is as disorganized as you were in high school]]></title>
            <link>https://medium.com/@yag3/health-insurance-cycle-is-as-disorganized-as-you-were-in-high-school-f2c879559485?source=rss-e93f994af386------2</link>
            <guid isPermaLink="false">https://medium.com/p/f2c879559485</guid>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[medical-insurance]]></category>
            <category><![CDATA[medical]]></category>
            <category><![CDATA[healthcare]]></category>
            <category><![CDATA[health-technology]]></category>
            <dc:creator><![CDATA[Yordanos Goshu]]></dc:creator>
            <pubDate>Thu, 20 Feb 2020 18:25:42 GMT</pubDate>
            <atom:updated>2020-02-20T18:25:42.907Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/637/1*VjyXPJ0A6R7JVzgWPTG_2g.jpeg" /></figure><h3>““If I had an hour to solve a problem I’d spend 55 minutes thinking about the problem and 5 minutes thinking about solutions.” — Yordanos Goshu” — Albert Einstein (or something like that)</h3><h3>Getting to your root</h3><p>Finding the root to a problem is inevitably where one finds the solution. How quickly one gets to the root is what leads to a strong product. There are many modes of de-risking your search that can be very pivotal. One of these modes with product development is customer discovery. This is the process of understanding whether the problem statement you and your team have concocted is actually a problem your users face, or are surface level analysis your team has made via a small number of observations.</p><h3>Our root</h3><p>Electronic Medical Records (EMR) are like the two folders that you used for all of your classes during your sophomore year of high school. You stuffed homework assignments, projects, class notes, and even your club flyers in there. Then exam season rolls around, and you scrummage through your folder trying to find the notes for your class. You inevitably: 1) lose important notes 2) can’t understand what you wrote 3) can’t organize your thoughts, etc. An insurance companies is like your equally disorganized friend in high school that asks to borrow your notes for the same exam. They try to parse information that you can barely understand to make an informed decision while taking an exam.</p><h3>Our Insight</h3><p>Our interviews have shown us that building a solution that tries to organize EMR systems is not solving our root problem. It, in actuality, is propagating the issue. The issue to be solved here is, referring to our high school analogy, buying more folders in the beginning of the school year (or at least having your dad force you to). These folders are a form of organization that allows us to anticipate required data structure down the health care insurance pipeline. If we know that doctors notes and electronic medical data is used for billing statements, and that there is currently an issue of ambiguity that causes providers to lose billions of dollars, why not better understand where the ambiguity arises from and buy new folder to properly organize it? *Intellimed has now joined the group chat*</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f2c879559485" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Insurer Dilemma: How Medical Billing is Being Disrupted by Deep Learning]]></title>
            <link>https://medium.com/@yag3/the-insurer-dilemma-how-medical-billing-is-being-disrupted-by-deep-learning-61cd8b59b94c?source=rss-e93f994af386------2</link>
            <guid isPermaLink="false">https://medium.com/p/61cd8b59b94c</guid>
            <category><![CDATA[natural-language-process]]></category>
            <category><![CDATA[young-adulthood]]></category>
            <category><![CDATA[healthcare]]></category>
            <category><![CDATA[medical-billing]]></category>
            <category><![CDATA[deep-learning]]></category>
            <dc:creator><![CDATA[Yordanos Goshu]]></dc:creator>
            <pubDate>Tue, 28 Jan 2020 01:33:37 GMT</pubDate>
            <atom:updated>2020-01-28T01:33:37.497Z</atom:updated>
            <content:encoded><![CDATA[<p>I believe that having to select your own health insurance should be a passageway to adulthood. After graduating, I went to my first employer and they gave me a list of three insurance providers, and I, more or less, blindly selected the “best” option (mainly due to all of the subconscious marketing delivered via my parents). My interpretation of this decision in math would be: if we define a random variable X which represents the number of doctor visits for a given person, what is the expected value of X given the person’s current health status. After a couple of Google searches and polling of friends, I couldn’t differentiate between the options my company provided. Fast forward a year and a half after that original selection, I’ve been to the doctor three times. Each time, I would receive the bill with three sections: “Service charge”, “Insurance covered”, “Payment Due” (or something along those lines). Like many, I see the “Payment Due:” section is ~zero and I ignore the bill and appreciate my insurance provider. Then the cycle continues.</p><p>This semester, at Cornell Tech, I began to explore the intricacies of how these bills are generated. Which is no easy task but that is where layers of abstractions will allow me to convey some key ideas that drive the relationship between health providers (e.g. hospitals) and insurance companies. During your doctor visit, you probably notice your doctor taking down notes as she asks you a set of questions. After you leave, the nurse and other health providers may also add notes to these notes depending on the recommendation of your doctor. These notes are then sent to a group of humans that read these notes and “code” the corresponding billing tag for each aspect of the doctors note. These bills are then sent to your insurance. After a few back-and-forth between the two, walla, you get your “Payment Due” section.</p><p>This task is quite interesting to me because at its root it exhibits features suited for knowledge based Natural Language Processing. These features include: 1) Imitating the coder’s ability to process the doctor’s notes (natural language) at scale. The corpora of the english language is what drives the scale issue here 2) Once these categories have been extracted, removing the ambiguity of the categories via relationship of these categories 3) Finally, there is a translation from these categories and their mapping to corresponding coding bills.</p><p>These billings drive a hundred billion dollar market, and mistakes arising due to ambiguity of translation (as described above) raise a serious issue. As engineers, we disambiguate via structure. In this area, being able to help the coders develop a deterministic process of coding is key. By integrating our statistical and knowledge based understanding of how doctors communicate, we will be able to extract key details from the notes. Specifically, given the advent of Deep Learning, we will be able to empirically estimate these statistical models giving us valuable interpretations of the doctor notes. Then we will move towards classification methods for determining relationships between the extracted categories (e.g. different form of kidney diseases). Finally, we will propose translation mechanisms to the billing code via the mappings defined by the insurer.</p><p>There are many exciting computer science problems presented in the above problem. Furthermore, there is great added value that comes from bringing transparency in the medical world. Transparency is key to improving the lives and well being of millions of Americans because it allows them to take ownership of their health. *IntelliMed has entered the conversation*</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=61cd8b59b94c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>