Lost in Time? A Meta-Learning Framework for Time-Shift-Tolerant Physiological Signal Transformation

Translating non-invasive signals such as photoplethysmography (PPG) and ballistocardiography (BCG) into clinically meaningful signals like arterial blood pressure (ABP) is vital for continuous, low-cost healthcare monitoring. However, temporal misalignment in multimodal signal transformation impairs waveform accuracy, especially in capturing critical features like ABP peaks. Conventional synchronization methods often rely on strict similarity assumptions or manual tuning, while existing Learning with Noisy Labels (LNL) approaches are ineffective against time-shifted supervision, either discarding excessive data or failing to correct label shifts. To address this, we propose ShiftSyncNet, a meta-learning-based bi-level optimization framework designed to automatically mitigate performance degradation due to time misalignment. It comprises a transformation network (TransNet) and a time-shift correction network (SyncNet), which learns time offsets between training pairs and applies Fourier phase shifts to align supervision signals. Experiments on one real-world industrial dataset and two public datasets show that ShiftSyncNet outperforms baselines by 9.4\%, 6.0\%, and 12.8\%, respectively. Results highlight its effectiveness in correcting time shifts, improving label quality, and enhancing transformation accuracy under diverse misalignments—attributable to SyncNet's label correction capabilities within meta-learning framework.