Interpretable Machine Learning for Modeling, Evaluating, and Refining Clinical Decision-Making
This thesis promotes the use of interpretable machine learning to model observed behavior, enabling the comparison of treatment strategies and quality assessment of policy evaluations. It explores representations of patient data that support interpretable modeling and proposes approaches that leverage structure in the data collection process to improve model interpretability. Finally, it argues that interpretable models of current behavior can inform the design of new policies.
Jul 28, 2025
