Memory-Guided Personalization for Physician-Specific Diagnostic Inference

Abstract

Recent advances in deep learning have improved diagnostic precision across medical imaging tasks, but clinical adoption remains limited because model outputs do not always align with diverse physician reasoning styles. We propose a physician-centric diagnostic framework for real-time adaptive inference tailored to individual clinicians. The framework consists of supervised learning, human-in-the-loop guidance, and personalized deployment. Physician feedback is encoded as memory-based priors and reused at inference without retraining, enabling lightweight end-to-end personalization. Experimental results on multiple medical imaging tasks demonstrate effective adaptation to individual diagnostic styles while maintaining strong accuracy.