Single-Beat Cuffless Blood Pressure Estimation Using Ear-PPG and ECG with a Lightweight Hybrid Learning Framework

Researchers proposed a hybrid ML framework for cuffless blood pressure estimation from single ear-PPG and ECG beats.

The model combines CNN-based beat embeddings with physiology-inspired features and LightGBM, using synchronized ear-PPG, ECG, and IMU data.

In tests on a small stress-protocol cohort and PulseDB, it achieved lower MAE than baseline methods.

The approach does not require multi-second signal windows, making wearable BP monitoring more practical and potentially more robust to motion and noise.