<div class="xblock xblock-public_view xblock-public_view-vertical" data-usage-id="block-v1:MITx+HST.953x+3T2020+type@vertical+block@913cec00700943bfa8e89f38fe6032e6" data-init="VerticalStudentView" data-graded="False" data-request-token="dca4b98043c411ef8f150e08775edbcd" data-block-type="vertical" data-runtime-version="1" data-course-id="course-v1:MITx+HST.953x+3T2020" data-has-score="False" data-runtime-class="LmsRuntime">
<h2 class="hd hd-2 unit-title">Introduction</h2>
<div class="vert-mod">
<div class="vert vert-0" data-id="block-v1:MITx+HST.953x+3T2020+type@html+block@9a3380dd718047538ce9ea733dc1ed35">
<div class="xblock xblock-public_view xblock-public_view-html xmodule_display xmodule_HtmlBlock" data-usage-id="block-v1:MITx+HST.953x+3T2020+type@html+block@9a3380dd718047538ce9ea733dc1ed35" data-init="XBlockToXModuleShim" data-graded="False" data-request-token="dca4b98043c411ef8f150e08775edbcd" data-block-type="html" data-runtime-version="1" data-course-id="course-v1:MITx+HST.953x+3T2020" data-has-score="False" data-runtime-class="LmsRuntime">
<script type="json/xblock-args" class="xblock-json-init-args">
{"xmodule-type": "HTMLModule"}
</script>
<h3>Section Introduction</h3>
<p>This part is subdivided into nine subsections that follow the common process of generating new medical evidence using clinical data mining. In <strong>2.01</strong>, the reader will learn how to transform a clinical question into a pertinent research question, which includes defining an appropriate study design and select the exposure and outcome of interest. In <strong>2.02</strong>, the researcher will learn how to define which patient population is most relevant for investigating the research question. Owing to the essential and often challenging aspect of analysis of EHRs, it will be described in the following four chapters elaborately. Subsections <strong>2.03</strong> and <strong>2.04</strong> deal with the essential task of data preparation and pre-processing, which is mandatory before any data can be fed into a statistical analysis tool. <strong>2.03</strong> explains how a database is structured, what type of data they can contain and how to extract the variables of interest using queries; <strong>2.04</strong> presents some common methods of data pre-processing, which usually implies cleaning, integrating, then reducing the data; <strong>2.05</strong> provides various methods for dealing with missing data; <strong>2.06</strong> discusses techniques to identify and handle outliers. In <strong>2.07</strong>, common methods for exploring the data are presented, both numerical and graphical. Exploration data analysis gives the researcher some invaluable insight into the features and potential issues of a dataset, and can help with generating further hypotheses. <strong>2.08 & 2.09</strong>, “data analysis”, presents the theory and methods for <strong>model development</strong> as well as common <strong>data analysis techniques</strong> in clinical studies, namely <strong>linear regression</strong>, <strong>logistic regression</strong> and <strong>survival analysis</strong> including Cox proportional hazards models. Finally, <strong>2.10</strong> discusses the principles of model validation and sensitivity analyses, where the results of a particular research are tested for robustness in the face of varying model assumptions.</p>
</div>
</div>
</div>
</div>
<div class="xblock xblock-public_view xblock-public_view-vertical" data-usage-id="block-v1:MITx+HST.953x+3T2020+type@vertical+block@60c1ec347f2c44deb5a27b897dc31b8f" data-init="VerticalStudentView" data-graded="False" data-request-token="dca4b98043c411ef8f150e08775edbcd" data-block-type="vertical" data-runtime-version="1" data-course-id="course-v1:MITx+HST.953x+3T2020" data-has-score="False" data-runtime-class="LmsRuntime">
<h2 class="hd hd-2 unit-title">Walkthrough</h2>
<div class="vert-mod">
<div class="vert vert-0" data-id="block-v1:MITx+HST.953x+3T2020+type@html+block@c4f23923439048808ddad3da57761723">
<div class="xblock xblock-public_view xblock-public_view-html xmodule_display xmodule_HtmlBlock" data-usage-id="block-v1:MITx+HST.953x+3T2020+type@html+block@c4f23923439048808ddad3da57761723" data-init="XBlockToXModuleShim" data-graded="False" data-request-token="dca4b98043c411ef8f150e08775edbcd" data-block-type="html" data-runtime-version="1" data-course-id="course-v1:MITx+HST.953x+3T2020" data-has-score="False" data-runtime-class="LmsRuntime">
<script type="json/xblock-args" class="xblock-json-init-args">
{"xmodule-type": "HTMLModule"}
</script>
<h3>Example: Indwelling Arterial Catheters</h3>
<p style="font-size: 16px;">This section includes worked examples inspired from a unique study, published in Chest in 2015 by Hsu et al., which addressed a key question in clinical practice in intensive care medicine: “is the placement of an indwelling arterial catheter (IAC) associated with reduced mortality, in patients who are mechanically ventilated but do not require vasopressor support?” IACs are used extensively in the intensive care unit for continuous monitoring of blood pressure and are thought to be more accurate and reliable than standard, non-invasive blood pressure monitoring. They also have the added benefit of allowing for easier arterial blood gas collection which can reduce the need for repeated arterial punctures. Given their invasive nature, however, IACs carry risks of bloodstream infection and vascular injury, so the evidence of a beneficial effect requires evaluation. The primary outcome of interest selected was 28-day mortality with secondary outcomes that included ICU and hospital length-of-stay, duration of mechanical ventilation, and mean number of blood gas measurements made. The authors identified the encounter-centric ‘arterial catheter placement’ as their exposure of interest and carried out a propensity score analysis to test the relationship between the exposure and outcomes using MIMIC. The result in this particular dataset (spoiler alert) is that the presence of an IAC is not associated with a difference in 28-day mortality, in hemodynamically stable patients who are mechanically ventilated. This case study provides a basic foundation to apply the above theory to a working example, and will give the reader first-hand perspective on various aspects of data mining and analytical techniques. This is in no way a comprehensive exploration of EHR analytics and, where the case lacks the necessary detail, we have attempted to include additional relevant information for common analytical techniques.</p>
<p style="font-size: 16px;">For the interested reader, the paper is available on <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4665738/" target="[object Object]">PubMed</a>.</p>
</div>
</div>
</div>
</div>
© All Rights Reserved