Client Session

ExamSoft Categories and Machine Learning to Identify Students at High Risk of Failing Board Exams







Session Description

Early identification of students most at risk of failing the COMLEX Level 1 board exam is an important step toward intervening on behalf of those students. Methods utilizing supervised machine learning approaches to predict the category to which something belongs (a spam e-mail vs. a legitimate e-mail for example) based upon some set of variables have become increasingly popular in recent years. Supervised machine learning requires a set of data with the outcome of interest labeled (for example success or failure), along with a set of variables to potentially predict such an outcome.

This session explores the use of ExamSoft categories and machine-learning methods to identify students most at risk of failing COMLEX Level 1 exams at an osteopathic medical school.



Matthew K. Pelletier, Ph.D.

Director of Assessment and Outcomes, Liberty University College of Osteopathic Medicine

Related Resources