Position: Beyond Prediction: Toward Verifiable Physiological Waveform Reasoning with Foundation Models and Agentic LLMs

Xiaoda Wang, Ching Chang, Defu Cao, Kaiqiao Han, Fang Sun, Yue Huang, Minxiao Wang, Chang Xu, Xiao Luo, Runze Yan, Xiangliang Zhang, Xiao Hu, Yan Liu, Yizhou Sun, Wei Wang, Carl Yang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:172441-172459, 2026.

Abstract

Physiological waveforms (e.g., ECG, PPG, EEG) encode clinically meaningful information in fine-grained morphology, precise timing, and cross-channel dynamics, yet most machine learning systems still treat them as generic time series and optimize end-to-end prediction. In this position paper, we argue for verifiable physiological waveform reasoning: extracting localized, measurable signal evidence from raw signals, interpreting that evidence into physiological semantics, and supporting clinically grounded decisions. Waveform reasoning is challenging due to acquisition heterogeneity, signal fidelity, complex semantics and cross-channel coupled dynamics. We analyze why existing model families remain insufficient: physiological foundation models learn strong perceptual representations but remain weak at verifiable reasoning, while LLM-based adaptations have limited waveform understanding. To bridge this gap, we advocate verifiable, closed-loop systems that unify waveform semantics with language intelligence. Concretely, we propose a dual-process architecture that System 1 aligns physiological waveforms with language, and System 2 provides agentic reasoning via a Plan–Act–Verify loop, together enabling verifiable physiological waveform reasoning. We further propose evaluations beyond accuracy, emphasizing traceability, replayability, counterfactual robustness, and calibrated abstention.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-wang26lz, title = {Position: Beyond Prediction: Toward Verifiable Physiological Waveform Reasoning with Foundation Models and Agentic {LLM}s}, author = {Wang, Xiaoda and Chang, Ching and Cao, Defu and Han, Kaiqiao and Sun, Fang and Huang, Yue and Wang, Minxiao and Xu, Chang and Luo, Xiao and Yan, Runze and Zhang, Xiangliang and Hu, Xiao and Liu, Yan and Sun, Yizhou and Wang, Wei and Yang, Carl}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {172441--172459}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/wang26lz/wang26lz.pdf}, url = {https://proceedings.mlr.press/v306/wang26lz.html}, abstract = {Physiological waveforms (e.g., ECG, PPG, EEG) encode clinically meaningful information in fine-grained morphology, precise timing, and cross-channel dynamics, yet most machine learning systems still treat them as generic time series and optimize end-to-end prediction. In this position paper, we argue for verifiable physiological waveform reasoning: extracting localized, measurable signal evidence from raw signals, interpreting that evidence into physiological semantics, and supporting clinically grounded decisions. Waveform reasoning is challenging due to acquisition heterogeneity, signal fidelity, complex semantics and cross-channel coupled dynamics. We analyze why existing model families remain insufficient: physiological foundation models learn strong perceptual representations but remain weak at verifiable reasoning, while LLM-based adaptations have limited waveform understanding. To bridge this gap, we advocate verifiable, closed-loop systems that unify waveform semantics with language intelligence. Concretely, we propose a dual-process architecture that System 1 aligns physiological waveforms with language, and System 2 provides agentic reasoning via a Plan–Act–Verify loop, together enabling verifiable physiological waveform reasoning. We further propose evaluations beyond accuracy, emphasizing traceability, replayability, counterfactual robustness, and calibrated abstention.} }
Endnote
%0 Conference Paper %T Position: Beyond Prediction: Toward Verifiable Physiological Waveform Reasoning with Foundation Models and Agentic LLMs %A Xiaoda Wang %A Ching Chang %A Defu Cao %A Kaiqiao Han %A Fang Sun %A Yue Huang %A Minxiao Wang %A Chang Xu %A Xiao Luo %A Runze Yan %A Xiangliang Zhang %A Xiao Hu %A Yan Liu %A Yizhou Sun %A Wei Wang %A Carl Yang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-wang26lz %I PMLR %P 172441--172459 %U https://proceedings.mlr.press/v306/wang26lz.html %V 306 %X Physiological waveforms (e.g., ECG, PPG, EEG) encode clinically meaningful information in fine-grained morphology, precise timing, and cross-channel dynamics, yet most machine learning systems still treat them as generic time series and optimize end-to-end prediction. In this position paper, we argue for verifiable physiological waveform reasoning: extracting localized, measurable signal evidence from raw signals, interpreting that evidence into physiological semantics, and supporting clinically grounded decisions. Waveform reasoning is challenging due to acquisition heterogeneity, signal fidelity, complex semantics and cross-channel coupled dynamics. We analyze why existing model families remain insufficient: physiological foundation models learn strong perceptual representations but remain weak at verifiable reasoning, while LLM-based adaptations have limited waveform understanding. To bridge this gap, we advocate verifiable, closed-loop systems that unify waveform semantics with language intelligence. Concretely, we propose a dual-process architecture that System 1 aligns physiological waveforms with language, and System 2 provides agentic reasoning via a Plan–Act–Verify loop, together enabling verifiable physiological waveform reasoning. We further propose evaluations beyond accuracy, emphasizing traceability, replayability, counterfactual robustness, and calibrated abstention.
APA
Wang, X., Chang, C., Cao, D., Han, K., Sun, F., Huang, Y., Wang, M., Xu, C., Luo, X., Yan, R., Zhang, X., Hu, X., Liu, Y., Sun, Y., Wang, W. & Yang, C.. (2026). Position: Beyond Prediction: Toward Verifiable Physiological Waveform Reasoning with Foundation Models and Agentic LLMs. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:172441-172459 Available from https://proceedings.mlr.press/v306/wang26lz.html.

Related Material