From Wearables to Warnings: Predicting Pain Spikes in Patients with Opioid Use Disorder

Proceedings of Machine Learning Research LEAVE UNSET:1–11, 2026 Conference on Health, Inference, and Learning (CHIL) 2026 From Wearables to Warnings: Predicting Pain Spikes in Patients with Opioid Use Disorder Abhay Goyal abhay@nimblemind.ai Nimblemind, USA Navin Kumar navin@nimblemind.ai Nimblemind, USA Kimberly DiMeola kdimeola@aptfoundation.org APT Foundation, New Haven, CT, USA Rafael Trujillo Florida International University, Miami, FL, USA Soorya Ram Shimgekar sooryas2@illinois.edu University of Illinois – Urbana Champaign Christian Poellabauer cpoellab@fiu.edu Florida International University, Miami, FL, USA Pi Zonooz pi@nimblemind.ai Nimblemind, USA Ermonda Gjoni-Markaj APT Foundation, New Haven, CT, USA Declan Barry declan.barry@yale.edu APT Foundation, Yale University School of Medicine, New Haven, CT, USA Lynn Madden lmadden@aptfoundation.org APT Foundation, Yale University School of Medicine, New Haven, CT, USA Abstract Chronic pain (CP) and opioid use disorder (OUD) are common and interrelated chronic medical con- ditions. Currently, there is a paucity of evidence- based integrated treatments for CP and OUD among individuals receiving medication for opioid use dis- order (MOUD). Wearable devices have the potential to monitor complex patient information and inform treatment development for persons with OUD and CP, including pain variability (e.g., exacerbations of pain or pain spikes) and clinical correlates (e.g., perceived stress). However, the application of large language models (LLMs) with wearable data for understanding pain spikes, remains unexplored. Consequently, the aim of this pilot study was to examine the clinical cor- relates of pain spikes using a range of AI approaches. We found that machine learning models achieved rela- tively high accuracy (>0.7) in predicting pain spikes, while LLMs were limited in providing insights on pain spikes. Real-time monitoring through wearable devices, combined with advanced AI models, could facilitate early detection of pain spikes and support personalized interventions that may help mitigate the risk of opioid relapse, improve adherence to MOUD, and enhance the integration of CP and OUD care. Given overall limited LLM performance, these find- ings highlight the need to develop LLMs which can provide actionable insights in the OUD/CP context. Data and Code Availability Data and code is available upon reasonable request. We collected data from 25 pa- tients with chronic pain and opioid use disorder from the APT Foundation. Institutional Review Board (IRB) The study was ap- proved by the Yale University Institutional Review Board (2000034714) and the APT Foundation Board of Direc- tors. © 2026 A. Goyal, N. Kumar, K. DiMeola, R. Trujillo, S.R. Shimgekar, C. Poellabauer, P. Zonooz, E. Gjoni-Markaj, D. Barry & L. Madden. arXiv:2511.19577v2 [cs.AI] 12 Jan 2026 Short Title 1. Introduction Chronic pain (CP) and opioid use disorder (OUD) are com- mon and interrelated chronic medical conditions. Chronic pain is defined as noncancer pain lasting most days for at least three months. The Centers for Disease Control and Prevention (CDC) estimates that 20% of adult Americans experience at least one lifetime episode of chronic pain and 8% have high-impact chronic pain that limits their life or work activities on most days Dahlhamer et al. (2016). Over two million individuals in the United States have OUD, which is a chronic relapsing condition that is asso- ciated with high rates of mortality Ahmad et al. (2023). Across the 12-month period ending in December 2024, the Centers for Disease Control and Prevention (CDC) re- ported an estimated 80,391 drug overdose deaths in the United States for Health Statistics (2025). Estimates suggest that at most 20% of adults with OUD are on MOUD, and 2019 adjusted estimates sug- gest past-year OUD affected over seven million people in the US Krawczyk et al. (2022). MOUD has shown sev- eral benefits such as decreases in mortality, increases in treatment adherence, decreases in heroin use, and aug- mented health, social and criminal justice outcomes Or- ganek (2022); Kumar et al. (2021); Mattick et al. (2009); Mun et al. (2019). Co-management of opioid use disorder and chronic pain is a growing clinical challenge: Preva- lence estimates of CP among patients on opioid agonist treatment (e.g., methadone or buprenorphine) are higher than the general population (37-61% vs. 20%) and may reflect reduced pain thresholds, increased pain sensitivity, or opioid-induced hyperalgesia Su (2025). Treatment of patients with CP and OUD presents challenges as unre- lieved pain is associated with illicit opioid and alcohol use, anxiety, and depression Su (2025), and OUD may in some cases develop as a result of opioid therapy for CP. Among patients with OUD, CP is associated with opi- oid cravings and pain exacerbation or spikes are linked to opioid use. However, chronic pain is not routinely as- sessed or addressed in opioid treatment programs,

This content is AI-processed based on open access ArXiv data.