MPT: A Comprehensive Toolbox for Deriving Cardiorespiratory Physiomarkers from Biosignals with LLM-Based Interpretation

Multimodal physiological waveforms are increasingly employed for real-time physiological assessment across clinical and ambulatory settings. However, existing open-source tools lack comprehensive support for analyzing multiple concurrent waveforms and the automated derivation of clinically meaningful physiomarkers with their interpretations. The Multimodal Physiology Toolbox (MPT) addresses this need through an integrated, publicly accessible web application built on a web-deployed software suite for extracting physiomarkers from electrocardiogram (ECG), photoplethysmogram (PPG), respiratory (RESP), and arterial blood pressure (ABP) signals. MPT facilitates advanced cardiorespiratory analysis with minimal user input and accommodates a wide spectrum of applications, from resting-state monitoring to intensive care recordings. MPT is the first toolbox that integrates physiomarker extraction with large language model (LLM)-based interpretation to generate concise, patient-specific physiological reports that contextualize key findings. Performance validation demonstrates the framework's generalizability and clinical applicability. An interactive MPT dashboard is publicly available at: https://mpt.kamaleswaran.com/.