M2AE (Biosignal Fingerprinting)

Summary

Cross-modal ECG↔PPG masked autoencoder foundation model (Oxford / SJTU). Learns compact modality-agnostic “biosignal fingerprints” from 3.4M+ paired 10 s segments; frozen fingerprints transfer to CVD, hypertension, mortality, demographics, and cross-modal reconstruction under missing-modality inference.

Datasets used