Skip to content

Predictive Modelling in the Health Sciences

Course Number
5200 (Biostatistics)
Course Instructor(s)
Konstantin Shestopaloff

Course Description

This course will introduce students to methods and approaches for predictive modelling and teach the associated technical knowledge and computational approaches required to successfully implement an applied analysis. The material covered will include the general approaches to predictive modelling and a selection of methods that would typically be used at various stages of a pipeline when training a model. This includes handling of missing data, variable selection and fitting a final predictive model. The approach of the course will be to familiarize the students with the methods and technicals via applications, with a focus on data that is commonly encountered in the health sciences.

Course Objectives

The main objectives of the course are to further refine knowledge of:

  • existing methods and approaches to be used for different types of data;
  • the stages of data processing and predictive modelling as relevant for health science data;
  • technical proficiency and the importance of efficiency in implementation.

Methods of Assessment

Assignments (4 @ 15% each; 2 @ 20% each) 100%
With the permission of the instructor:
Data analysis project
Plus, best three best assignments (@ 15% each)

General Requirements

Intro/mid-level statistics (ANOVA, linear regression, logistic regression, model diagnostics) or mathematics (calculus, linear algebra), basic programming skills.

Recommended: An applied statistics course at the senior level (STA1001H/1002H/1008H), predictive modelling experience (CHL5229H/5230H; CSC411H1) or equivalent knowledge; Basic computer science (CSC148H1 or equivalent); Knowledge of R or S+.