AI-ENABLED DIGITAL HEALTH ASSISTANT FOR EARLY DETECTION AND MANAGEMENT OF OSTEORADIONECROSIS IN HEAD AND NECK CANCER PATIENTS: PROPOSED CLINICAL TRIAL DESIGN

Abstract

Background: Head and neck cancers (HNC) are prevalent in South Asia, particularly in Pakistan, where radiotherapy remains a cornerstone of treatment. Osteoradionecrosis (ORN) of the mandible is a serious late complication of radiation, leading to pain, trismus, infection, and impaired quality of life [1,2]. Early detection of ORN is critical but challenging in low-resource settings.

Objective: To evaluate the efficacy of an AI-enabled smartphone application (“OsteoAware”) in the early detection and management of ORN among patients receiving curative-intent radiotherapy for HNC.

Methods: We designed a parallel-group, randomized controlled trial enrolling 200 HNC patients at two tertiary centers in Pakistan. Participants are randomized 1:1 to receive either standard care or standard care plus the OsteoAware app, which uses patient-reported symptoms and a machine learning model to deliver targeted self-care prompts and alerts for clinician referral. The follow-up period is 12 months.

Primary Outcome:Time-to-diagnosis of ORN.
Secondary Outcomes:ORN incidence and grade, symptom burden (EORTC QLQ-H&N35), pain severity, quality of life (EORTC QLQ-C30), and app usability metrics.

Preliminary Results (): Of 200 patients, ORN developed in 5% in the app group versus 12% in controls (p<0.05). Median time to ORN diagnosis was shorter with app use (4.5 vs. 7.8 months). Pain and xerostomia scores were significantly lower, and patient-reported quality of life was higher in the app arm.

Conclusion:If validated in this setting, OsteoAware may enable early ORN detection and improve outcomes in HNC patients in LMICs.