Estimating Clinical Lab Test Result Trajectories from PPG using Physiological Foundation Model and Patient-Aware State Space Model -- a UNIPHY+ Approach

2025-09-19

AI Summary

This study addresses the clinical challenge of intermittent and invasive blood sampling, which creates blind spots in biochemical dynamic monitoring. We propose UNIPHY+Lab, the first framework enabling long-term, non-invasive estimation of multiple biochemical trends directly from continuous, non-invasive photoplethysmography (PPG) signals. Methodologically, it integrates a large-scale PPG foundation model (for local waveform encoding) with a FiLM-modulated Mamba architecture (for long-range temporal modeling); FiLM dynamically adapts the initial state to patient-specific physiological baselines and enables multi-task joint biomarker prediction. Evaluated on two ICU datasets, UNIPHY+Lab significantly outperforms LSTM and forward-filling baselines: for most biomarkers, MAE and RMSE decrease by over 20%, and R² improves substantially. These results validate the feasibility and advancement of physiology-guided, patient-specific PPG modeling for continuous, non-invasive biochemical monitoring.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Clinical laboratory tests provide essential biochemical measurements for diagnosis and treatment, but are limited by intermittent and invasive sampling. In contrast, photoplethysmogram (PPG) is a non-invasive, continuously recorded signal in intensive care units (ICUs) that reflects cardiovascular dynamics and can serve as a proxy for latent physiological changes. We propose UNIPHY+Lab, a framework that combines a large-scale PPG foundation model for local waveform encoding with a patient-aware Mamba model for long-range temporal modeling. Our architecture addresses three challenges: (1) capturing extended temporal trends in laboratory values, (2) accounting for patient-specific baseline variation via FiLM-modulated initial states, and (3) performing multi-task estimation for interrelated biomarkers. We evaluate our method on the two ICU datasets for predicting the five key laboratory tests. The results show substantial improvements over the LSTM and carry-forward baselines in MAE, RMSE, and $R^2$ among most of the estimation targets. This work demonstrates the feasibility of continuous, personalized lab value estimation from routine PPG monitoring, offering a pathway toward non-invasive biochemical surveillance in critical care.

Problem

• Estimating clinical lab test trajectories from non-invasive PPG signals
• Capturing extended temporal trends in laboratory values over time
• Performing personalized multi-task estimation for interrelated biomarkers

Innovation

• PPG foundation model for waveform encoding
• Patient-aware Mamba for temporal modeling
• FiLM-modulated states for personalization