Publication

Abstract : Early detection is one of the most effective ways to save lives from breast cancer, but current screening methods such as annual mammography can miss cases in women with dense breast tissue and cannot track changes between appointments. Customized, real-time screening solutions are being made possible by advancements in wearable biosensors, medical imaging, and artificial intelligence (AI). In this paper, we introduce a framework that brings these technologies together: a comfortable wearable device that continuously monitors bio signals, a deep learning model that analyzes mammographic images, and a patient-specific digital twin that simulates breast health over time. The wearable records indicators such as skin temperature, electrical impedance, and heart rate variability. Combined with imaging data, these signals are processed by a ResNet-50–based AI model to detect subtle, early changes linked to cancer risk. The digital twin updates continuously, forecasting individual risk and recommending screening schedules tailored to each person rather than fixed intervals. Validation using the Curated Breast Imaging Subset of the Digital Database for Screening Mammography (CBIS-DDSM) showed high sensitivity for cancer detection. By reducing false negatives and supporting earlier clinical decisions, this approach could complement radiologist-led diagnosis, shorten delays, and make screening more personalized especially for women who are not well served by existing methods.