Physics-Constrained Deep Learning Model for Contactless Blood Pressure Monitoring from Triaxial Bodyseismography
A research paper proposes Phy-BP, a non-invasive blood pressure estimation framework using triaxial bodyseismography (BSG). It includes an adaptive quality-control algorithm and a physical model of 3D wave propagation embedded into a deep learning model. Experiments were conducted on a 162-hour hospital dataset from 21 subjects.
The paper introduces Phy-BP, a framework for contactless blood pressure monitoring using triaxial bodyseismography. It addresses challenges in traditional ballistocardiography by aligning multi-axis features through a physics-constrained deep learning model. The approach was validated on a 162-hour hospital dataset collected from 21 subjects.
The integration of a physical wave propagation model into deep learning aligns triaxial BSG signals, potentially improving robustness against body-bed interaction variations. The adaptive quality-control algorithm selects cardiogenic-rich segments, which may enhance signal fidelity.
This research could advance unobtrusive long-term blood pressure monitoring, relevant to remote patient monitoring and wearable health devices. The use of hospital data suggests a path toward clinical validation.
If validated, the technology could enable new products in continuous, contactless cardiovascular monitoring for hospitals, home care, and consumer health markets.
Next signals include peer-reviewed publication, replication on larger diverse datasets, and potential commercialization or clinical trials. Watch for follow-up studies on generalizability across different bed types and patient populations.