Problem definition: Effective hypertension management is critical to reducing consequences of atherosclerotic cardiovascular disease, a leading cause of death in the United States. Clinical guidelines for hypertension can be enhanced using decision-analytic approaches, capable of capturing many complexities in treatment planning. However, model-generated recommendations may be uninterpretable/unintuitive, limiting their acceptability in practice. We address this challenge by investigating interpretable treatment planning.
Methodology/results: We study interpretable treatment planning for Markov Decision Processes (MDPs) and specifically analyze the problems of optimizing monotone policies, which increase treatment intensity with worsening patient health, and optimizing class-ordered monotone policies (CMPs), which generalize monotone policies by imposing monotonicity over state and action classes rather than states and actions. We establish that both policies depend on initial state distributions. Furthermore, optimal monotone policies can be generated tractably for many treatment planning problems. Next, we propose exact formulations for optimizing interpretable policies broadly. Then, we define and analyze the price of interpretability (PI), proving that the CMP’s PI does not exceed the monotone policy’s. Finally, we formulate, parameterize, and evaluate MDPs for 10-year hypertension treatment planning using a large, nationally representative dataset of the United States population. We compare the structure and performance of optimal monotone policies and CMPs to optimal MDP-based policies and current clinical guidelines. At the patient-level, optimal MDP-based policies may be unintuitive, recommending more aggressive treatment for healthier patients than sicker patients. Conversely, monotone policies and CMPs never de-escalate treatment, reflecting clinical intuition. Across 66.5 million patients, optimized monotone policies and CMPs outperform clinical guidelines, saving over 3,246 quality-adjusted life years per 100,000 patients, while paying low PIs. Sensitivity analysis further reveals that monotone policies and CMPs are robust to various definitions of “interpretability.”
Managerial implications: Interpretable policies can be tractably optimized, drastically outperform existing guidelines, and pay low PIs – potentially increasing the acceptability of decision-analytic approaches in practice.
G.-G. P. Garcia, L. N. Steimle, W. J. Marrero, and J. B. Sussman (2022). Interpretable Policies and the Price of Interpretability in Hypertension Treatment Planning. Optimization Online [Pre-print].