21010 - ADVANCED STATISTICS FOR HEALTH SCIENCES
Department of Decision Sciences
Course taught in English
Course Director:
MARCO BONETTI
MARCO BONETTI
Mission & Content Summary
MISSION
This course will provide the foundational elements of statistical theory, methods and models that are useful in the health sciences. The principles discussed will form the backbone for later developments in the ML and AI courses. Problems of estimation will be dealt with within the finite sample size setting and the large sample size setting. The R software will be used to discuss and appreciate the properties of the procedures, and to implement the methods.
CONTENT SUMMARY
- Review of classical inference: Random variables and statistics, point estimation, confidence intervals, hypothesis testing.
- Approximate distributions and large-sample inference. Maximum likelihood estimators via numerical maximization.
- The linear model.
- The ANOVA model for comparing the means of two or more groups. Intro to multiple hypothesis testing and adjustments. Sample size determination for two-group problem.
- Measures of association. Contingency tables, risk difference, risk ratio, odds ratio, Simpson’s paradox.
- Generalized Linear models. Logistic regression for classification.
- Fisher's Linear Discriminant Function. Binary diagnostic testing (Bayes theorem), ROC and AUC.
- Nonparametric tests.
- Survival Analysis: Estimators of the survival funcion, testing. Regression models.
- Frailty and shared frailty models for risk prediction.
Intended Learning Outcomes (ILO)
KNOWLEDGE AND UNDERSTANDING
At the end of the course student will be able to...
- Understand the key guiding principles and concepts of statistical inference.
- Build a toolkit of methods and models.
- Prepare to profitably attend courses on advanced topics in Machine Learning and Artificial Intelligence.
APPLYING KNOWLEDGE AND UNDERSTANDING
At the end of the course student will be able to...
- Formulate suitable probabilistic models for statistical problems in the health sciences.
- Apply appropriate statistical methodology to estimation and testing problems.
- Gain experience in the practical implementation of the methods and models.
- Interpret the outcome of the estimation procedure in view of the application at hand.
Teaching methods
- Lectures
- Practical Exercises
DETAILS
- Face-to-face lectures. Theory, exercises, and some R will be blended during the lectures.
- PC-based labs ("Bring Your Own Laptop"). Labs will be devoted to exercises and hands-on sessions to appreciate the concepts and work on examples, also using the R programming language.
Assessment methods
| Continuous assessment | Partial exams | General exam | |
|---|---|---|---|
|
x |
ATTENDING AND NOT ATTENDING STUDENTS
- A 2-hour in-class, closed book and closed notes exam will be held after the course, within each of the three exam sessions.
- The exam will be handwritten on paper.
- Some of the questions (for a total of 21 points out of 31) will cover the methods, models, and properties discussed in class, to assess the understanding of the theoretical aspects.
- Some of the questions (for a total of 10 points out of 31) will require the analysis of some data using R. Students will bring their own laptop to the exam and will not be required to submit any of the code or output: only the final answers of the analyses and the associated comments will need to be reported on the paper. No access to the internet will be allowed during the exam.
- There will be no difference in assessment method / exam program between attending and non attending students.
Teaching materials
ATTENDING AND NOT ATTENDING STUDENTS
- Materials for the course will be distributed through the Blackboard platform. They will consist of class notes, R code, and additional handouts. For more details on the topics we suggest the following volumes:
- Dalgaard P (2008). Introductory Statistics with R. Springer, New York, NY.
- Dobson AJ and Barnett AG (2018). An Introduction to Generalized Linear Models, Fourth Edition. Chapman & Hall / CRC, Boca Raton, FL.
- Lawless J (2003). Statistical Models and Methods for Lifetime Data. Wiley, New York, NY.
Last change 03/06/2025 17:46