RADIOMICS AND MACHINE LEARNING FOR NEUROPATHIC PAIN SEVERITY PREDICTION IN HEAD AND NECK CANCER SURVIVORS

Abstract

Neuropathic pain is a persistent and clinically challenging complication among head and neck cancer survivors, often resulting from tumor invasion, surgery, radiotherapy, chemotherapy, or combined treatment-related nerve injury. Accurate prediction of neuropathic pain severity is essential for early risk stratification, personalized pain management, and improved survivorship care. This study, titled “Neuropa-Radiomics and Machine Learning–Based Prediction of Neuropathic Pain Severity in Head and Neck Cancer Survivors,” proposes an integrated predictive framework that combines radiomic imaging features with clinical and treatment-related variables to estimate neuropathic pain severity. Radiomic features extracted from head and neck imaging were analyzed alongside patient-level characteristics, including cancer site, treatment modality, radiation exposure, surgical history, and pain-related clinical indicators. Machine learning models were developed to identify complex patterns associated with mild, moderate, and severe neuropathic pain outcomes. The proposed approach aims to support objective pain severity prediction by capturing imaging-based tissue changes and their relationship with nerve injury and post-treatment pain burden. By integrating radiomics with machine learning, this study highlights the potential of data-driven decision support systems in oncology pain management. The findings may assist clinicians in identifying high-risk survivors, optimizing follow-up strategies, and designing individualized interventions to reduce long-term neuropathic pain and improve quality of life.