Beyond Prediction: Toward Verifiable Physiological Waveform Reasoning: Advances, Challenges, and Future Directions

Résumé fourni par la source

Physiological waveforms (e.g., ECG, PPG, EEG) are high-fidelity measurements of underlying biological dynamics. Meaningful clinical interpretation hinges on localized morphology, precise timing, cross-lead consistency, and signal-quality awareness. Over the past few years, the field has seen rapid progress. However, most systems remain prediction-centric: they output labels or fluent rationales without exposing the localized, measurable evidence that supports intermediate claims and final decisions. This tutorial advocates verifiable physiological waveform reasoning, where conclusions are supported by explicit evidence objects. Following the theme of Advances, Challenges, and Future Directions, we (i) review recent advances in physiological foundation models and LLM-centric adaptations, (ii) analyze key limitations and waveform-specific reasoning challenges, and (iii) outline future directions toward verifiable, closed-loop systems with clearer reasoning objectives, structures, and evaluation. We highlight a dual-process blueprint and a Verification Ledger that enable evidence-grounded, replayable assessment beyond accuracy, including traceability, robustness, and calibrated abstention.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.