Machine Learning Applications for Opioid Use Management in Chronic Cancer Pain: A Systematic Scoping Review

Chronic pain remains a critical clinical issue worldwide, with adverse effects on the quality of life of oncology patients. Meanwhile, the overuse of opioids to treat or alleviate chronic cancer pain has contributed to a global opioid crisis. The increasing accessibility of high-quality clinical datasets and computational frameworks has promoted the use of machine learning (ML) techniques in clinical practice to manage opioid consumption. This review investigates the current bibliography referring to the role of applied ML techniques in opioid administration in patients with chronic cancer pain. The objective of the current scoping review, according to population, intervention, comparison, and outcome (PICO) standards, was to evaluate the effectiveness of ML techniques in monitoring opioid consumption in patients with chronic cancer pain. This review includes scientific journal papers published from 2010 to 2024 that use healthcare data from patients with chronic cancer pain, apply machine learning techniques, and may address the potential consequences of the misuse of opioids. A systematic literature search, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, was performed in PubMed and Google Scholar databases. Data extracted include the study’s goal, dataset used, cohort selected, types of ML models created, model evaluation metrics, and the details of the ML tools and techniques used to create the models. After conducting the screening process, 50 articles were identified, but only four focused specifically on or included data of patients with chronic cancer pain where ML techniques were applied. The four included studies showed high performance (area under the curve {AUC}: >0.8) in predicting opioid adherence, misuse, and long-term use. Although generalizability remains limited due to small sample sizes and a lack of external validation, it sets distinct limits in applying these methods in clinical use. After a thorough review of recent literature, ML models demonstrated promising accuracy in predicting opioid adherence, misuse, and long-term use among patients with chronic cancer pain. However, these findings are based on studies with limited sample sizes and a lack of external validation, which restricts their generalizability. Future research should focus specifically on populations with chronic cancer pain and expand predictive models to incorporate a combination of clinical, psychosocial, biometric, and genomic data. This approach may enable more accurate, personalized, and safer opioid management in oncology care.