PhD student: Margaux Törnqvist
Title: A Causal Machine Learning Approach to Personalized Antidepressant Prescription in Major Depressive Disorder
Supervisors: Pr Cédric Lemogne, Dr Chloé Geoffroy
Doctoral school: ED393
Promotion: 2025
Funding: funded by Theremia & ANRT
Thesis abstract
Depression is one of the leading causes of disability worldwide. Many antidepressants are available, yet clinical guidelines offer limited guidance on how to choose between them. In practice, prescribing therefore relies largely on a trial-and-error approach, even though response to antidepressants varies considerably between patients. This PhD explores the hypothesis that this heterogeneity could be explained by individual patient characteristics, paving the way for more personalised prescribing. To this end, we leverage large-scale real-world data, which reflect prescribing practices in routine care. As such data are prone to confounding bias, we use causal machine learning methods, which can estimate treatment effects while limiting this bias and exploring their heterogeneity. The ultimate goal is to identify patient profiles that respond differently to antidepressants, in order to better inform their prescription.
