Share Course Ware
Health Sciences > Microbiology > Medical Decision Support
 Medical Decision Support  posted by  duggu   on 11/27/2007  Add Courseware to favorites Add To Favorites  
Further Reading
More Options

Ohno-Machado, Lucila, and Staal Vinterbo, HST.951J Medical Decision Support, Fall 2005. (Massachusetts Institute of Technology: MIT OpenCourseWare), (Accessed 09 Jul, 2010). License: Creative Commons BY-NC-SA

Comparison of logistic regression vs. neural networks as prognostic models.

Comparison of logistic regression vs. neural networks as prognostic models. (Image by Prof. Lucila Ohno-Machado.)

Course Highlights

In addition to sample exams, a number of lecture notes and homework assignments are available for this course.

Course Description

This course presents the main concepts of decision analysis, artificial intelligence and predictive model construction and evaluation in the specific context of medical applications. It emphasizes the advantages and disadvantages of using these methods in real-world systems and provides hands-on experience. Its technical focus is on decision support, knowledge-based systems (qualitative and quantitative), learning systems (including logistic regression, classification trees, neural networks, rough sets), and techniques to evaluate the performance of such systems. It reviews computer-based diagnosis, planning and monitoring of therapeutic interventions. It also discusses implemented medical applications and the software tools used in their construction. Students produce a final project using the machine learning methods learned in the course, based on actual clinical data.

*Some translations represent previous versions of courses.



Help support MIT OpenCourseWare by shopping at! Partnering with, MIT OCW offers direct links to purchase the books cited in this course. Click on the book titles and purchase the book from, and MIT OCW will receive up to 10% of all purchases you make. Your support will enable MIT to continue offering open access to MIT courses.

Who Should Take this Course?

The course is required for students in the Master's Program in Medical Informatics, but is open to other graduate students and advanced undergraduates.


Below are some examples of projects developed in this course, by domain:

  1. Anonymization of Databases
    • Using Boolean Reasoning to Anonymize Databases
  2. Diagnostic Models
    • Using Patient-Reportable Clinical History Factors to Predict Myocardial Infarction
    • A Genetic Algorithm to Select Variables in Logistic Regression: Example in the Domain of Myocardial Infarction
  3. Prognostic Models
    • Development and Evaluation of Models to Predict Death and Myocardial Infarction Following Coronary Angioplasty and Stenting
    • Major Complications after Angioplasty in Patients with Chronic Renal Failure: A Comparison of Predictive Models

There is no required textbook for this course, however, the following textbooks are recommended for reference:

Hastie, T., R. Tibshirani, and J. H. Friedman. The Elements of Statistical Learning: Data Mining, Inference, and Prediction. (Springer Series in Statistics.) New York, NY: Springer Verlag, October 2001. ISBN: 0387952845.

Duda, Richard O., Peter E. Hart, and David G. Stork. Pattern Classification. 2nd ed. New York, NY: John Wiley & Sons, November 2000. ISBN: 0471056693.


Familiarity with SAS and MATLAB® will be helpful, and consultation with respective user manuals may be necessary.


30% Homework

Homework's are due at the end of each module. They may require programming, and all code must be included. They are to be solved individually. Homework's received after the deadline may be subject to substantial grade penalty. No homework's will be accepted after the solutions have been handled.

30% Midterm

The midterm will contain topics from decision analysis and machine learning. There will be a strict time limit of 1.5 h. Students should bring class notes, homework solutions, and readings.

40% Final Project

The final project has to be developed for this course and should be done individually. Although the project may contain parts developed previously for another purpose, it is essential that substantial effort be demonstrated into developing a project specifically for this class. A final report in the form of a 5+ page paper is expected, as well as a demonstration of the implementation in the form of a 15 minute presentation. Students are advised to discuss their projects with the instructors ahead of time. Examples of final reports will be handled during the course.

MATLAB® is a trademark of The MathWorks, Inc.




  LEC #       TOPICS
  1       Introduction to Medical Decision Support
  2       Simple Probabilistic Reasoning
  3       Fuzzy & Rough Sets - Part 1
  4       Fuzzy & Rough Sets - Part 2
  5       Bayesian Networks - Part 1: Representation & Reasoning
  6       Bayesian Networks - Part 2: Learning From Data
  7       Logistic Regression - Part 1
  8       Logistic Regression - Part 2
  9       Unsupervised Learning
  10       Classification Trees & CART
  11       Artificial Neural Networks
  12       Support Vector Machines
  13       Evaluation of Predictive Models - Part 1
  14       Evaluation of Predictive Models - Part 2
  15       Optimization and Complexity
  16       Survival Analysis
  17       Review of Predictive Methods
          Midterm Exam
  18       Review of Complexity
  19       Applied Informatics in Cardiology
  20       Review of Clustering
  21       Student Presentations
  22       Student Presentations
  23       Student Presentations   Tell A Friend