Introduction to Machine Learning for Pharmacoepidemiologists

Course Aim
The aim of this course is to provide background and guidance for pharmacoepidemiologists around the use of machine learning (ML) and in considering potential projects with an ML component. This course will help attendees understand how ML can be utilized in the context of pharmacoepidemiology, map terminology used in informatics to one commonly used in pharmacoepidemiology, and review differences in the analytic approach compared to traditional statistical modelling. Key categories of ML will be reviewed with examples of algorithms for each. Finally, the course will cover basic data preparation and tools required for a ML analysis. An interactive exercise will be included to reinforce key concepts and demonstrate practical applications of techniques discussed through the course. The course faculty aims to promote engagement between the pharmacoepidemiology community and ML/informatics community.

Requisites Statement
Level of expertise: entry level
· This course will assume limited/no previous exposure to ML
· This course is intended to help pharmacoepidemiologists have a basic understanding of machine learning techniques to aid in the development of collaborative projects
· A basic understanding of epidemiology is required

Course Objectives
· To understand how and when ML can be used for pharmacoepidemiology research
· To learn common ML terminology and understand how these terms differ from standard pharmacoepidemiology terminology
· To provide a high-level overview of the steps in a project involving ML, including design considerations, data preparation, modelling, model validation, and interpretation of results
· To review real-world examples of ML use in pharmacoepidemiologic research
· To participate in an interactive exercise applying key considerations learned during the course

Syllabus Outline
· Introduction
· Session 1: The applications of machine learning in pharmacoepidemiology
· Session 2: When to use ML in pharmacoepidemiology
· Session 3: Types of machine learning and associated algorithms
· Session 4: Supervised machine learning workflow: from data to deployment
· Session 5: Patient finding: Using ML to address underdiagnosis
· Session 6: De-mystifying deep learning
· Interactive session: Putting it all into practice
· Q&A