<?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 FrankdeGraaf on Medium]]></title>
        <description><![CDATA[Stories by FrankdeGraaf on Medium]]></description>
        <link>https://medium.com/@frankdegraaf_48347?source=rss-6c341569dbb1------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Stories by FrankdeGraaf on Medium</title>
            <link>https://medium.com/@frankdegraaf_48347?source=rss-6c341569dbb1------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 21:43:28 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@frankdegraaf_48347/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[Travel bugs and data science. A great opportunity for biomarker discovery?]]></title>
            <link>https://medium.com/orikami-blog/travel-bugs-and-data-science-a-great-opportunity-for-biomarker-discovery-1fab50cc9f3f?source=rss-6c341569dbb1------2</link>
            <guid isPermaLink="false">https://medium.com/p/1fab50cc9f3f</guid>
            <category><![CDATA[quantified-self]]></category>
            <category><![CDATA[biomarker-discovery]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[wearables]]></category>
            <category><![CDATA[digital-biomarkers]]></category>
            <dc:creator><![CDATA[FrankdeGraaf]]></dc:creator>
            <pubDate>Tue, 12 May 2020 07:01:00 GMT</pubDate>
            <atom:updated>2020-06-09T07:42:13.895Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*D7voCISyBgay6Pch3XC18g.png" /></figure><p><em>I am a typical example of someone with the travel bugs. While I am travelling, I always feel more relaxed and more at ease. Even when something stressful happens, e.g. not being able to find a place to sleep, it does not get me upset. Whenever I get home I try to stay in this relaxed state, but I fail every time again. This got me thinking, is it possible to quantify this relaxed state of mind? Can I identify measurable factors that represent or influence this state? In other words, is it possible to discover a </em><a href="https://blog.orikami.nl/the-birth-of-a-digital-biomarker-a4a829370201"><em>digital biomarker</em></a><em> for this mental well-being?</em></p><p><em>And there it was. An opportunity to explore this thought popped up October last year. During a 3-month long trip I used my wearable which I had already been using a few months at home. To what extend is a digital biomarker plausible by using this wearable data? The journey begins! In this blog I will guide you in this initial exploration. Subsequently the opportunity at hand, the data collection and some initial hypothesis testing are discussed.</em></p><blockquote><a href="https://blog.orikami.nl/what-we-learned-from-creating-digital-biomarkers-for-fatigue-a6a330cb1ca0">Digital biomarkers</a> are measurable, quantifiable indicators for a biological condition or state, obtained from connected and smart technology.</blockquote><h4><strong>Part 1: The opportunity at hand</strong></h4><p>Ever since I was I little child I have been fascinated by traveling, but the travel bugs never left me after my trip to south America in 2014. I remember the feeling that hit me when I was hiking in National Park Torres del Paine and seeing glacier Grey: my first glacier. The feeling of freedom and the feeling of being so small on such a big, beautiful planet. It was priceless. Traveling keeps you moving and enables you to put things in perspective. About a year ago the desire to travel started acting up again and last October I decided to go for another 3-month trip to Nepal and the Philippines.</p><p>A few months before, I had become interested in measuring biometrical data from myself. It started with optimizing my training schedule for the running competition <em>MS motion</em>. I bought a sport watch and used it to develop- and follow a specific training schedule. Whilst seeing all the data that the sport watch collected, I became curious about learning more about myself using this personal, quantified data. It is the dream of any data scientist: getting lots of data on yourself on which you can perform all kinds of analysis to gain insights. I became an example of a trend called ‘<a href="https://www.ted.com/talks/gary_wolf_the_quantified_self?language=en">the quantified self</a>’.</p><blockquote>The <a href="https://www.ted.com/talks/gary_wolf_the_quantified_self?language=en">quantified self</a> is a trend in which you get self-knowledge through numbers and is sometimes also called life-logging. It has been over 10 years that the term ‘quantified self’ has been introduced by Gary Wolf and Kevin Kelly.</blockquote><p>The dots were connected. The combination of my quantified self and the planned trip provided an even greater opportunity to gain insights. I now got access to substantial amounts of data for two different, relevant situations: (1) ‘the being at home, working Frank’ and (2) ‘the more relaxed, traveling Frank’. By comparing the data for these two situations, the relevance of the measured parameters for my feeling of a relaxed state of mind can be explored. In other words, this gives insights about these factors’ relevance for a digital biomarker for this type of mental well-being.</p><p>Why do we want to know this? Good question! First, I would like to be in this relaxed state of mind as much as possible. By having a digital biomarker for this state, I can experiment with altering my behaviour and analyse the corresponding effect on this state. Furthermore, the identified metrics might give some clues to what are influencing factors I can experiment with. So, I rubbed my hands and started exploring.</p><h4><strong>Part 2: Data collection and cleaning</strong></h4><p>Well, let us get more specific about the data collection and cleaning to see what parameters and how much data we are talking about. Are you all about details? Keep on reading!</p><p><strong><em>2a. Data collection<br></em></strong>I used the sport watch <a href="https://support.garmin.com/en-US/?productID=571520&amp;tab=topics">Garmin VivoActive3</a> for the data collection. This watch, as most wearables, uses photoplethysmography (PPG) to measure heart rate. A fancy name, but what it basically means is that this technique shines light into the skin and measures the light scattered by the blood flow. This way it indirectly detects the heart rate. This watch thereby contains Inertial Measurement Units (IMUs) to detect movement (accelerometer, gyroscope, and magnetometer). Garmin developed algorithms to translate the raw data of these sensors in different parameters, of which the most interesting four are listed in Table 1. As you can see in this table, a stress parameter is already calculated. This will be our first suspect for the relaxed state of mind- biomarker.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jes1aUk2azE4ACUsSrykGg.png" /><figcaption>Table 1. List of parameters measured by the Garmin VivoActive3</figcaption></figure><p>The watch saves- and is able to export the data in two different ways: (1) the raw *.FIT files and (2) the aggregated data per specific time period in *.csv format. For this first exploration I started with the aggregated data with 1 day as time period. This meant manual downloads of *.csv files for each day and for each parameter <em>or,</em> as I did, build a small program to do that for you. So, I built a web scraper (based on the <a href="https://selenium-python.readthedocs.io/">Selenium library</a>) that gathered all data and recombined it into one dataframe. To this dataframe I added a column with the label whether I was at ‘Home’ or ‘Traveling’ and a column with the label indicating in which Country I was. As an example, the first three rows of this dataframe are depicted in Table 2.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VQzyIa894URSvI0o0Gvb4g.png" /><figcaption>Table 2. First rows of the resulting dataframe after data collection and recombination</figcaption></figure><p><strong>2b. Data cleaning<br></strong>Data cleaning is needed to get rid of invalid data that is a consequence of incomplete measurements. I won’t get into the extensive details here, but I performed data cleaning for the following two occasions. First, I need to filter out the data of the days that I did not wear my watch the entire day since that data is incomplete. Secondly, due to measurement accuracy of the RHR- and STRESS data, the data should be removed when there has been less than 4 hours of sleep on that day (see Table 1).</p><p>Figure 1 shows the amount of data points per parameter per country that are left after the data cleaning. Clearly more data is present from the home situation, however I still have &gt;60 days of measurements for each parameter from the traveling situation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*LWS5l9WRBPHf574MlMPHUA.png" /><figcaption>Figure 1. Number of data points per parameter per Country.</figcaption></figure><p>As an example, Figure 2 shows what the data over time looks like for the parameter “Steps” where the different country labels are shown by the background colour.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wSVF-Sft_cDnfYqGwaxJHQ.png" /><figcaption>Figure 2. The number of steps measured by the Garmin VivoActive3 from July 2019 until March 2020. Background colour represents in which country these points are measured. The Netherlands is the home country, Nepal and the Philippines are travels.</figcaption></figure><blockquote><strong>Example: The importance of data quality<br></strong>Looking at Figure 2, you see that the number of steps per day is relatively high in the first weeks of Nepal. The reason: I was hiking around Mount Everest for the first 17 days. However, the highest number of steps was the 18th of November in the Philippines with ~28000 steps. I specifically remember this day, because while I was travelling by scooter (!) I got messages from my watch all the time that I achieved my step goal for the day again. This is an<strong> invalid data point</strong>. Invalid points should be filtered out as much as possible, otherwise invalid conclusions can be made. One should thus be careful with data interpretation when data quality is not optimal!</blockquote><h4><strong>Part 3: Hypothesis testing</strong></h4><p>Are you still with me? Now the highlight of the tour begins: the exploration of the data to see if<strong> </strong>there are any relations between the different measurements and the fact that I’m at home or travelling (i.e. in Nepal or in the Philippines). The labels ‘Home’ and ‘Traveling’ thereby also respectively represent the ‘Non- relaxed state of mind’ and ‘relaxed state of mind’. This is of course a simplification but let’s start with this. Furthermore, for this first exploration, I presume that the data in different countries are independent samples and the distributions for the different labels can be compared. Table 3 lists the hypotheses that I would like to test.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*D8Qx5JQek7UhThPYPjUXeg.png" /><figcaption>Table 3. Hypotheses to test between data collected at home and data collected while traveling</figcaption></figure><p>As an example, Figure 3 shows the distributions of our number 1 suspect for the relaxed state of mind: ‘Stress level’. To see whether these distributions of stress level are significantly different, statistical comparison tests are performed with a significance level of alpha =0.05. For all four hypotheses the distributions are first tested for normality with a Shapiro-Wilcoxon test. As a result, the comparison of the distributions is performed by a one-tailed- independent t-test and Mann-Whitney test for respectively normal and non-normal distributions. Table 4 shows the resulting p-values for all 4 hypotheses.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xhsSDkfRyqHXGDlSmCbopQ.png" /><figcaption>Figure 3. The distributions of measured stress level for the labels: ‘Home’ and ‘Travel’</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FRNM0LBjh51qVv_oOZwAoQ.png" /><figcaption>Table 4. Outcome of the statistical tests for the 4 hypotheses stated in table 3.</figcaption></figure><p>What does this all mean? Let’s explain it one by one:</p><ol><li>The built-in stress level measurement by Garmin is statistically different between ‘Home’ and ‘Travel’. However, it gives higher values when traveling than when at home. This does not seem to coincide with my feeling of the relaxed state of mind, thus this stress parameter does not seem to be an appropriate measure.</li><li>The heartrate at rest is a good second suspect as a factor of my relaxed state of mind. However, the heartrate at rest while traveling and while being at home are not statistically significant different.</li><li>The number of steps is statistically significant larger when I am traveling. This is also what I would expect since back home I mostly sit behind a desk and while traveling I am out exploring. Increased physical activity possibly contributes to my relaxed state of mind. I do notice this too when I am at home and do physical exercise. A good parameter to investigate further.</li><li>The amount of sleeping hours is not statistically different between being at home and traveling. This is unexpected: I actually expected a larger number of sleeping hours while traveling. This is a case in which the objective data shows a different story than I would have thought based on my gut/memory. This shows that being able to collect this data offers the ability to make more precise conclusions. Even though the amount of sleep is not different, I still expect that <a href="https://www.tandfonline.com/doi/abs/10.3109/10253890.2015.1087503">sleep is a contributing factor</a> to a relaxed state of mind. Possibly it is not the amount of sleep, but the quality of the sleep.</li></ol><p>A lot more questions arise when exploring and analysing the data, but let’s stop for now. In the box below, some limitations of the current analysis and ideas for future work are given.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*u1sM_yJJ4lH1OXerAuedxg.png" /></figure><h3><strong>Closing words</strong></h3><p>I hope you liked my data science journey. At least I did. Let’s look back at the objective. The goal was to discover factors representing or influencing a biomarker for the relaxed state of mind that I have while travelling. For this first exploration I looked at the aggregated data per day of four different parameters as measured by my Garmin VivoActive3. The distributions of these parameter values for the ‘Home’ and ‘Travel’ situation are compared statistically. The analysis done so far was a first exploration and there are some limitations to the performed analysis. However, some clues about the relevance of the measured parameters are obtained:</p><ul><li>The stress parameter as calculated by Garmin is significantly higher when I am traveling than when I’m at home and thus does not seem to represent my feeling of a relaxed state of mind.</li><li>The heartrate at rest <em>and</em> sleeping hours show no significant difference between ‘Home’ and ‘Travel’, indicating that these are not representative factors for the relaxed state of mind. Especially the zero difference in number of sleeping hours was surprising!</li><li>The number of steps, representing the amount of physical action is a lot higher when traveling than when at home. This is an indication that the number of steps (representing physical action) might be a factor for having a relaxed state of mind.</li></ul><p>Many questions are still present, and the data offers a lot more opportunities for future work. For now, going through this process though I hope to have given you as reader an idea about:</p><ol><li>The discovery of parameters for a digital biomarker</li><li>Data exploration in a quantified self-situation and…</li><li>How the combination of travel bugs and data science gives interesting opportunities!</li></ol><p>It has been a pleasure to write this while at the same time looking at data about myself. I want to challenge you to think along with me, so please share your comments and suggestions below!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1fab50cc9f3f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/orikami-blog/travel-bugs-and-data-science-a-great-opportunity-for-biomarker-discovery-1fab50cc9f3f">Travel bugs and data science. A great opportunity for biomarker discovery?</a> was originally published in <a href="https://medium.com/orikami-blog">Orikami blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Using data in healthcare is hot: how to define a strategy not to burn your fingers!?]]></title>
            <link>https://medium.com/orikami-blog/using-data-in-healthcare-is-hot-how-to-define-a-strategy-not-to-burn-your-fingers-90b167620e3e?source=rss-6c341569dbb1------2</link>
            <guid isPermaLink="false">https://medium.com/p/90b167620e3e</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[healthcare]]></category>
            <category><![CDATA[strategy]]></category>
            <dc:creator><![CDATA[FrankdeGraaf]]></dc:creator>
            <pubDate>Wed, 19 Dec 2018 10:13:23 GMT</pubDate>
            <atom:updated>2018-12-21T08:16:27.860Z</atom:updated>
            <content:encoded><![CDATA[<p><em>How to find the added value of using data (science) for healthcare</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/619/1*D-0rUqxq_MzqER4sCSW1vg.png" /><figcaption>Illustration by Chloë Vervoort, Orikami ©</figcaption></figure><h3>Introduction</h3><p>Healthcare slowly starts to realize the enormous potential of data. A thousand reasons exist why data is so interesting for healthcare, but one reason should be enough here: <strong>healthcare costs are continuously increasing, and this is simply not sustainable</strong>. To illustrate this: as a ratio of GDP, the government health expenditures in Europe went from 6.2% in 2002 to 7.1% in 2016<a href="#_ftn1">[1]</a>. Healthcare, therefore, must become more efficient and one of the ways is by using data. A great example is the Portuguese hospital Sao Joao by implementing the HVITAL system. The HVITAL system collects physiological signs of every patient in the hospital and analyses this real-time to identify and give alerts regarding patients at risk for clinical deterioration<a href="#_ftn2">[2]</a>. This way the doctor or nurse receives a message when additional assessment or treatment is needed and care is given when- and where it is necessary.</p><p>But how do you get to these kinds of data solutions? What problems can you solve by using data? How do you know the best way to extract value from data and what data do you actually need? It is important to realize that data is always a <em>means</em> and not a <em>goal in itself. </em>To successfully become more data-driven, a coherent data strategy is needed that describes the purpose of using data and how it can bring added value. This data strategy guides you in knowing what to aim for and how to accomplish this step by step. In this blog, I’ll give you some insights on how to determine a good data strategy by a process we at Orikami call <em>Data Value Mapping</em>.</p><h3>Data strategy in healthcare</h3><p>A data strategy is a strategy for organizing, governing, analyzing and deploying an organization&#39;s information assets.<a href="#_ftn3">[3]</a> In other words, it is a strategy describing how data can have added value for your organization and how this added value can be obtained. In any organization, numerous possibilities of using data are possible. A good data strategy prioritizes these possibilities and shows you the way to go. The data strategy is best visualized in a so-called data roadmap, showing the steps you must undertake to go from the current situation to the future (data-driven) version of your organization.</p><p>To give you some ideas of how data can provide added value in healthcare, some examples are given below for three different segments within the healthcare domain. I’ll elaborate on these possibilities in a future blog. In the end, all examples come down to making your organization smarter by using data leading either to increased cost efficiency, better care or preferably the combination of both <a href="#_ftn4">[4]</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/983/1*vNfGB1j7mH2cr2vuSBGFog.png" /><figcaption>Examples of the added value of data (science) for three different domains in healthcare</figcaption></figure><h3>How to determine a good data strategy in healthcare?</h3><p>In the <em>Data Value Mapping</em> process, we map the (potential) added value of current- and future data sources on the strategy and vision of the organization. This intertwinement of the data strategy and the company strategy is essential to achieve long-term constructive results. The data value mapping process consists of several steps which are schematically visualized in the figure below. In the following section I´ll guide you through the different steps of a data value mapping and within the quotation marks a specific example of a data value mapping we did for <em>Hersenz</em>. Hersenz is a treatment program, given by 13 different institutes in the Netherlands, for people with acquired brain injuries such as a cerebrovascular accident.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*db7jL1h0kXHAxNg0bYskZg.png" /><figcaption>Schematic visualization of the Data Value Mapping process of Orikami ©</figcaption></figure><p><strong>1.</strong> <strong>Exploration of current and future data sources</strong></p><p>Before starting anything new, start by looking at what you already have. You’ll be surprised by how many data sources are already present in your organization and it would be a waste to not use these. In order to see their (potential) value, it is important to assess these data sources on accessibility, quantity, and quality. The quality and consistency of your data determine the quality and consistency of the conclusions you will draw from your data, so don’t underestimate this step. Explorative trend- and correlation analysis can thereby help to see the potential value of this data. To get a complete picture of data possibilities for your data strategy it is essential to do the same assessment for other possible data sources. For example, you can think of the use of wearables to generate data.</p><blockquote>The most important data source at Hersenz is the Routine Outcome Monitoring (ROM) database. This database contains the results from five different questionnaires that are periodically administered to their clients. In this first step, an explorative analysis on this ROM data is performed to see the amount of data available per client, how many missing or wrong values are present and how consistent the actual data is with the acquisition protocol. The potential value of this ROM data is assessed by looking at the treatment effect for different parameters and correlations analysis of these outcomes with gender and age.</blockquote><p><strong>2. Define company strategy</strong></p><p>Data solutions need to contribute to the goals that your organization has, e.g. to achieve the best possible care with reduced costs. One, therefore, needs to define the organization&#39;s strategy before a good data strategy can be built. Brainstorm about the ideal future state of your organization <a href="#_ftn5">[5]</a>. Due to the ever-changing interference of the payers in healthcare, it is important to also assess your business model. Will it stay as it is or do you foresee a different model for the future? A second step is to think about the current situation, for example in the form of a <em>Strengths Weaknesses Opportunities Threats (SWOT) </em>analysis <a href="#_ftn6">[6]</a>. A stakeholder mapping, for example by a power interest matrix <a href="#_ftn7">[7]</a>, can also provide valuable insights giving direction to your data strategy.</p><blockquote>Where does Hersenz want to stand in 2030? Together with a few representatives from Hersenz we did a brainstorm session in the form of a cover story[5] to find the answer to this question. One of the headlines in this vision was that the positive effect of the Hersenz treatment on the long-term quality of life had been proven.</blockquote><p><strong>3.</strong> <strong>Combining and explore (the “actual” <em>Data Value Mapping</em>)</strong></p><p>The actual Data value mapping is to map the possibilities of current- and future data sources to accomplish the company’s strategic goals. In our experience, you need to do this in a co-creation process in which data experts, company leaders, and domain experts are present. The outcome of this mapping is a list of possible uses of data with their potential value. The SWOT analysis can be used to prioritize the abundance of ideas to find the biggest opportunities.</p><blockquote>By connecting the current data sources and possible future data sources to Hersenz’ vision for 2030 many possibilities arose. For example, Hersenz’ treatment can be optimised by using the ROM data on the effect of the treatment in a real-time feedback loop to the practitioner to better monitor progress and adjust the treatment for the individual client. Another example is that Hersenz´ can benefit from using new data sources such as digital biomarkers and wearables to provide real world evidence towards payers and being able to show long term effects.</blockquote><p><strong>4.</strong> <strong>Development of the data roadmap</strong></p><p>Develop a data roadmap showing the most promising ways of using data and which steps need to be done at what moment to become more data-driven and a smarter organization.</p><blockquote>For Hersenz, all the possibilities discovered at step 3 were prioritised based on their potential in added value and amount of work that is yet needed. The most promising idea was worked out in concrete steps that Hersenz must take from now until 2025. Most important for Hersenz is the generation of evidence on the effect of their treatment, both for the payers and for themselves to be able to improve the treatment. Orikami gave advice on how to improve the current data acquisition and how to use the current- and other future data sources in the future to optimally generate evidence Hersenz is able to use towards payers and for treatment optimisation.</blockquote><h3>What to do with the data strategy — data roadmap?</h3><p>So now we have a data strategy in the form of a data roadmap, what’s next? Having a data roadmap means you have thought about all the possible ways to extract value from data and formulated the biggest opportunities and what work needs to be done. Next step is the actual doing! Work out the most promising ideas in proof- of concepts and, if successful, implement them as full products. By doing this in an iterative fashion you can quickly move on once something turns out to be unsuccessful and keep implementing the successful proofs-of-concept at the same time.</p><blockquote><em>Note: Don’t forget to occasionally revise your data strategy! Just as your business strategy, it might change because of the ever-changing environment.</em></blockquote><h3>To summarize…</h3><p>Data in healthcare offers great potential for offering better- and more efficient care, for example in the form of personalization. If you don’t want to end up with random data products with limited use- and limited added value, it is essential to start by making a good data strategy. A way to do this is by following our 4-step <em>Data Value Mapping </em>process<em>. </em>Once you have developed your data roadmap you can start extracting value from data step-by-step!</p><p>If you want more information about data strategy in healthcare or if you require help to develop data solutions for healthcare, please don’t hesitate to contact me at <a href="/service/mailto:frankdegraaf@orikami.nl">frankdegraaf@orikami.nl</a> or look at our site <a href="http://www.orikami.nl/">www.orikami.nl</a></p><h3>References</h3><p><a href="#_ftnref1">[1]</a> <a href="https://ec.europa.eu/eurostat/statistics-explained/index.php/Government_expenditure_on_health">https://ec.europa.eu/eurostat/statistics-explained/index.php/Government_expenditure_on_health</a></p><p><a href="#_ftnref2">[2]</a> Almeida, J. P. (2016). A disruptive Big data approach to leverage the efficiency in management and clinical decision support in a Hospital. <em>Porto Biomedical Journal</em>, <em>1</em>(1), 40–42. <a href="https://doi.org/https://doi.org/10.1016/j.pbj.2015.12.001">https://doi.org/https://doi.org/10.1016/j.pbj.2015.12.001</a></p><p><a href="#_ftnref3">[3]</a> <a href="https://hbr.org/2017/05/whats-your-data-strategy">https://hbr.org/2017/05/whats-your-data-strategy</a></p><p><a href="#_ftnref4">[4]</a> <a href="https://blog.orikami.nl/the-birth-of-a-digital-biomarker-a4a829370201">https://blog.orikami.nl/the-birth-of-a-digital-biomarker-a4a829370201</a></p><p><a href="#_ftnref5">[5]</a> <a href="https://gamestorming.com/cover-story/">https://gamestorming.com/cover-story/</a></p><p><a href="#_ftnref6">[6]</a> <a href="http://mci.ei.columbia.edu/files/2012/12/An-Essential-Guide-to-SWOT-Analysis.pdf">http://mci.ei.columbia.edu/files/2012/12/An-Essential-Guide-to-SWOT-Analysis.pdf</a></p><p><a href="#_ftnref7">[7]</a> <a href="https://www.researchgate.net/figure/stakeholder-mapping-the-power-versus-interest-grid-the-gird-shows-stakeholders-on-a_fig2_275894050">Https://www.researchgate.net/figure/stakeholder-mapping-the-power-versus-interest-grid-the-gird-shows-stakeholders-on-a_fig2_275894050</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=90b167620e3e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/orikami-blog/using-data-in-healthcare-is-hot-how-to-define-a-strategy-not-to-burn-your-fingers-90b167620e3e">Using data in healthcare is hot: how to define a strategy not to burn your fingers!?</a> was originally published in <a href="https://medium.com/orikami-blog">Orikami blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>