Researchers have developed OraRefer, a novel deep learning model designed for oral cancer screening using smartphone photographs. Unlike previous models that assess individual images, OraRefer fuses multiple photographs of a patient, incorporating risk factors to make a patient-level referral decision. This approach aims to improve sensitivity by ensuring that a minimal percentage of patients with disease are missed. The model demonstrated strong performance on a dataset of 3000 photographs from 714 patients, achieving an AUROC of 0.919 and a 95% detection rate for potentially malignant disorders or cancer at a 90% sensitivity target. AI
IMPACT This research could lead to more accessible and accurate early detection of oral cancer through widely available smartphones.
RANK_REASON Academic paper detailing a new deep learning model for medical screening. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
