AI-Driven Prediction of Cancer Pain Episodes: A Hybrid Decision Support Approach
Yipeng Zhuang, Yifeng Guo, Yuewen Li, Yuheng Wu, Philip Leung-Ho Yu, Tingting Song, Zhiyong Wang, Kunzhong Zhou, Weifang Wang, Li Zhuang
TL;DR
The study addresses forecasting breakthrough cancer pain in hospitalized lung cancer patients by introducing a hybrid AI workflow that couples time-aware structured ML with a retrieval-augmented LLM to leverage unstructured notes. Using a retrospective cohort of 266 patients, the approach integrates dynamic analgesic dosing features and clinical narratives to predict pain at 48 and 72 hours, achieving high accuracy and improved sensitivity through LLM augmentation. The methodology combines rigorous preprocessing, multi-model ML evaluation, and a carefully engineered LLM prompt and fusion strategy to produce clinically interpretable predictions. The results demonstrate potential for proactive analgesia, better resource allocation, and scalable integration into oncology workflows.
Abstract
Lung cancer patients frequently experience breakthrough pain episodes, with up to 91% requiring timely intervention. To enable proactive pain management, we propose a hybrid machine learning and large language model pipeline that predicts pain episodes within 48 and 72 hours of hospitalization using both structured and unstructured electronic health record data. A retrospective cohort of 266 inpatients was analyzed, with features including demographics, tumor stage, vital signs, and WHO-tiered analgesic use. The machine learning module captured temporal medication trends, while the large language model interpreted ambiguous dosing records and free-text clinical notes. Integrating these modalities improved sensitivity and interpretability. Our framework achieved an accuracy of 0.874 (48h) and 0.917 (72h), with an improvement in sensitivity of 8.6% and 10.4% due to the augmentation of large language model. This hybrid approach offers a clinically interpretable and scalable tool for early pain episode forecasting, with potential to enhance treatment precision and optimize resource allocation in oncology care.