Personalized Treatment for Opioid Use Disorder

Abstract

Wearable devices have the potential to revolutionize treatments for opioid use disorder (OUD) by measuring patient responses to different treatment regimens in real-time, enabling the development of personalized treatments. A variety of wearable devices with different features, sensitivities, and costs are available. Whether such devices are practical and cost-effective to incorporate in treatments for OUD, and if so how they should be used, are critical questions. To investigate these questions, we build a finite-horizon, non-stationary constrained partially observable Markov decision process (CPOMDP). To facilitate the solution of our model, we provide a novel budget reformulation that finds all optimal solutions lying on the original formulation’s solution convex hull. We then show that our reformulation can be solved using a binary search in conjunction with an exact POMDP algorithm. Applying those elements and using parameters estimated from past literature, we perform a numerical study to investigate the values of different wearables in OUDtreatments, where we consider different levels of budget, wearable precision, and patient treatment adherence(TA). We find that wearables can be valuable when patients react differently to treatments across the entire population. Furthermore, this value is the largest at low or moderate budgets for patients with low or moderate TA. Outside of these settings, the marginal benefit of wearables is negligible relative to their cost.