Forehead BCG Signal Analysis: Characterization and Robust Detection of Heartbeats Using Cosine Similarity and Energy-Based Approach

This work presents a ballistocardiogram (BCG) based beat-to-beat heart rate (HR) measurement algorithm for a portable that on the forehead. To address the diversity of BCG signals, we propose a beat-to-beat detection algorithm that can adapt to the varying shapes of BCG signals among different users. This algorithm utilizes similarity in shape, energy, and variations in interbeat intervals (IBI) to locate the heartbeat points. The algorithm has been verified on 10 subjects, where BCG and ECG signals were simultaneously recorded. By referencing the ECG signal, we have achieved a mean absolute relative error of 3.843% for beat-to-beat interval estimation. Moreover, on average, 99.442% of the beat-to-beat intervals are accurately detected.