Adaptive Evidential Meta-Learning with Hyper-Conditioned Priors for Calibrated ECG Personalisation

Keywords: ECG personalisation, evidential deep learning, uncertainty calibration, meta-learning, hyper-networks, class-conditional priors, few-shot adaptation, clinical deployment

TL;DR: We propose an evidential meta-learning framework with hyper-conditioned priors for well-calibrated ECG model personalisation from few-shot patient data.

Abstract: This research addresses a fundamental gap in uncertainty calibration during ECG model personalisation. We propose \emph{Adaptive Evidential Meta-Learning}, a framework that attaches a lightweight evidential head with hyper-network-conditioned priors to a frozen ECG foundation model. The hyper-network dynamically sets the evidential prior using robust, class-conditional statistics computed from a few patient-specific ECG samples. Trained via a two-stage meta-curriculum, our approach enables rapid adaptation with well-calibrated uncertainty estimates, making it highly applicable for real-world clinical deployment where both prediction accuracy and uncertainty awareness are crucial.

Submission Number: 111