Early Prediction of Work Disability in SLE Patients: Using Machine Learning to Model Labor Force Dropout to Guide Early Clinical Intervention
Date:
Javier Mencia Ledo, Dr. Gary Collins, Dr. George Tomlinson, Dr. Zahi Touma, Dr. Nowrouzi-Kia
It was a privilege to present my research at the Bombardier Research Rounds, where I engaged with clinicians and researchers from the Arthritis Program at UHN (University Health Network).
My talk, titled “Early Prediction of Work Disability in SLE Patients: Using Machine Learning to Model Labor Force Dropout to Guide Early Clinical Intervention,” focused on shifting the clinical approach to work disability from reactive management to proactive risk stratification. During the session, I shared our progress in creating a functional profile of work disability specifically for patients with Systemic Lupus Erythematosus (SLE), with the goal of informing the development of multidisciplinary, evidence-based interventions.
I reviewed the methodological framework used to build our risk stratification model, emphasizing how we leveraged longitudinal clinical data to detect latent signals of labor force dropout. A key highlight of the discussion was our estimation of the “warning window”—identifying exactly how far in advance of permanent disability onset a patient’s data triggers a high-risk flag. We also explored the specific clinical and socioeconomic factors driving these risk score fluctuations, providing the audience with a transparent look at the “why” behind the predictions to better support early clinical decision-making.
Click SLErisk.pdf to view it