Healthcare Databases & Information Systems Course


Topic: Review & Reporting Data

Learning Module Objectives:

After completing the activities in this section, you should be able to:

  • Review course content
  • Plan for activities before and after presentations
  • Present findings from your data analysis

Learning Material

  1. Revisiting course overview.  Take a look at what was promised at start of the course.  Now in hindsight, did the course deliver on its promises? Slides Video
  2. How to present high-dimensional data to clinicians Video► Slides►

Teach One Assignment

There is no teach one assignment for this week.

Individual Assignments

No individual assignment should be completed in teams.  Submit the URL for your work in Blackboard.    Describe your work predictive model for diabetes using narrated slides posted to a public domain such as You Tube.  If you do not wish to post the work to a public domain, discuss the issue with your instructor so an alternative could be provided.  The alternative will still require you to go through the steps of posting but not make the document public. 

Question Related to Review of the Course:  In this exercise, students ask and answer questions about topics in the course from and for each other.  Use the discussion group to ask at least 1 question. You can select any topic in the course that still remains confusing.  Do not ask administrative questions such as “When is the exam?” or "Where is the office of the instructor?". Ask only about what remains confusing. In asking a question, specify the topic and provide as much detail so that other students can answer the question.  Take time to write a good question. Start the question with a reference to the topic, such as:

"In the topic ‘Basic SQL Command’, in question 3, I get a different answer than other students.  Here is my code, could you help me understand what I am doing wrong? Provide snippet of your code.  Provide only the relevant portion  …."

After you have asked your question(s), go back and provide answers to one or more questions asked by others. A question can have multiple answers, so you can answer a question that has already been addressed by someone else.  Answer questions that remain unanswered, have in your view an incomplete or incorrect answer, or the answer provided is not as efficient as your solution. In providing the answer, give as much detail as possible. If you are suggesting a new coding approach, provide the code. If you wish you can provide a URL to a brief video.  The instructor will review the answers and 4 students who have answered more correct and more complete answers will receive 10 point bonus in their grade for teach one assignments. 

Question Related to Reporting Data: Suppose you have a presentation planned.  Suppose you have circulated your report to the people expected to be at the presentation.  You have not yet made the presentation.  Should you also call up people expected to be at your presentation and ask for their comments before you make your presentation?  It may take a great deal of time to call various decision makers and they may not have read your report (presumably you can provide them with a synopsis on the phone), is it really necessary to call and ask for their feedback? 

Ongoing Project Question: This week you finish the ongoing individual project with a final report.  Your presentation should have the following separate slides and each slide should not have an accompanying recording that exceeds 2 sentences:

  1. Title Slide:  This slide should have a brief title with font size of 40 to 60 points.  It should contain your name or alias.  It should have a statement in font size 12 that this presentation was prepared for your class at Health Informatics program at George Mason University.  
  2. Importance Slide: This slide should describe why is it important to screen for diabetes using medical history of the patient, as presented in the electronic health record.  Here you can refer to percent of diabetic patients not diagnosed.  Get this from PubMed. 
  3. Source & Size of Data Slide: This slide presents the size of the data and the recording describes the source of the data.  Describe the data as simulated data from a major health care institution in united states.  Show in this slide the number of unique patients, the number of records on these patients.
  4. Definition Slide: Describe how diabetes was defined.
  5. Predictors Slide: describe that a comprehensive set of diagnoses were used to predict the presence of diabetes or the risk for diabetes.  Give the total number of diagnoses for which you had at least 29 observations.  Explain how age and gender was added to these predictors.  Do not exceed 2-3 sentences in your recordings.  
  6. Obvious Predictors Slide:  Give examples of obvious predictions (these are diagnoses that always occur only with diabetes or have the word diabetes in their name but were not used in the definition of diabetes).
  7. Rare Disease Slide: Give example of 10 rare diseases that are predictive (these are diagnoses that occur more than 29 times but less than other diagnoses)
  8. Important Predictors Slide: Give examples of 10 most predictive diagnoses that are not obvious
  9. Data Mining Method Slide: Describe how naive Bayes was used to aggregate data from individual diagnoses to overall risk for diabetes. 
  10. Example Slide: Give an example of a case with 5 diagnoses, corresponding likelihood ratio, and calculated odds of diabetes.
  11. Cross-Validation Slide:  Describe how data was set aside for cross validation.  Explain how predictors that occur after diabetes were ignored. 
  12. Accuracy Slide:  Report the cross-validated accuracy of the predictions using receiver operating curves.  Describe the meaning of the curve in one or two sentences only.
  13. Next Step:  Summarize if the tool can be used to screen for diabetes.  Speak to the value of EHR-based screening using predictive models. 

Please do not use any of the titles given in the list above in your slides.  In general, slides do not need titles if the graphics or words in the slide are self explanatory.  Each slide should have a minimal set of words.  Stay with font size of 32 or more.  Do not use color indiscriminately.  If you use images from the web, indicate the source and make sure that it is not copyright protected.

Team Assignment

There are no team assignments this week.

Copyright © Farrokh Alemi, Ph.D. First created on January 9th 2005. Most recent revision 08/29/2018. This page is part of the course on Healthcare Databases.