RESEARCH | AI/ML
PCOS Diagnosis
& Artificial Intelligence
DMSc Thesis | Northeastern University | 2026
THESIS SUBMITTED | PUBLICATION IN PROGRESS
THE CLINICAL PROBLEM
PCOS is the most common endocrine disorder affecting reproductive-aged women, yet up to 70% of those who have it remain undiagnosed. The average time from symptom onset to diagnosis is over two years. This thesis asks why, and what a more rigorous diagnostic system might look like.
THE RESEARCH
This work examines the current diagnostic landscape for PCOS, evaluates existing AI and machine learning approaches to diagnosis, and identifies a critical structural limitation in how those models are built and trained. It proposes a framework for doing better.
A core methodological commitment throughout is thinking carefully about causality before modeling, making sure the right questions are being asked before reaching for technical solutions.