Title : BiotwinXAI Eye: A consumer centric digital twin for personalized explainable prediction of diabetic retinopathy and glaucoma risks
Abstract:
Diabetic retinopathy (DR) and glaucoma cause pre ventable blindness in over 600 million people worldwide. Early detection is critical, but challenging due to limited specialist access. We present BioTwinXAI-Eye, a digital twin system that creates a virtual model of each patient’s eye health by combining retinal images, eye pressure data, OCT scans, and health indi cators from consumer devices such as smartwatches and glucose monitors. Our system uses intelligent algorithms that track how diseases progress over time, and how different eye structures relate to each other. To help doctors trust the system, we include tools that explain predictions by highlighting important image regions and identifying key risk factors. Tests on 88,702 retinal images and 15,234 glaucoma patients show that our method improves accuracy by 21% compared to existing approaches. Crucially, it increased doctor confidence in AI predictions by 58% through transparent explanations. BioTwinXAI-Eye enables personalized care by continuously adapting to each patient’s unique health profile, supporting modern precision medicine goals in preventing vision loss.
