AI-assisted prediction of oral potentially malignant disorder transformation using serial smartphone images

Abstract

Oral potentially malignant disorders (OPMDs) represent a diverse group of mucosal lesions with varying risks of progression to oral squamous cell carcinoma, making timely detection and continuous surveillance essential for improving patient outcomes. Conventional follow-up relies on periodic clinical examinations, which may fail to capture subtle morphological changes occurring between visits. The widespread availability of smartphone cameras has created new opportunities for longitudinal image-based monitoring, enabling patients and clinicians to document lesion evolution conveniently and cost-effectively. Artificial intelligence (AI), particularly deep learning and computer vision techniques, offers the capability to analyze serial smartphone images, identify changes in lesion characteristics, and estimate the likelihood of malignant transformation with greater objectivity and consistency. AI-assisted systems can evaluate variations in lesion size, color, texture, border irregularity, and surface characteristics while providing automated risk stratification to support clinical decision-making. Integration with telemedicine platforms further expands access to specialist evaluation, especially in underserved and remote communities. Despite promising advancements, challenges remain regarding image standardization, dataset diversity, model generalizability, patient privacy, and clinical validation. Future developments focusing on explainable AI, multimodal data integration, and standardized imaging protocols are expected to enhance diagnostic reliability and facilitate personalized surveillance strategies. AI-assisted serial smartphone imaging has the potential to become a valuable adjunct for early detection, risk prediction, and long-term management of OPMDs.