Continuous Cuffless Blood Pressure Estimation via Effective and Efficient Broad Learning Model.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Cardiovascular Health

Background:

• Hypertension is a critical risk factor for cardiovascular diseases and mortality.
• Accessible blood pressure (BP) measurement is vital for hypertension management.
• Current deep learning models (DLMs) for cuffless BP estimation face computational challenges.

Purpose of the Study:

• To introduce an end-to-end broad learning model (BLM) for efficient cuffless BP estimation.
• To reduce computational complexity and enhance training efficiency compared to DLMs.
• To explore an incremental learning mode for memory efficiency and flexibility.

Main Methods:

• Developed a broad learning model (BLM) that increases network width over depth.
• Implemented an incremental learning mode (IBLM) for horizontal scalability and memory efficiency.
• Validated the model on a large dataset (403.67 hours) from the UCI database.

Main Results:

• The standard BLM (SBLM) achieved a mean absolute error (MAE) of 11.72 mmHg for arterial BP (ABP) waveform estimation, comparable to DLMs.
• SBLM demonstrated a 25.20-fold improvement in training efficiency over DLMs like LSTM and 1D-CNN.
• SBLM achieved MAE values of 3.04 mmHg for systolic BP (SBP) and 2.57 mmHg for diastolic BP (DBP).

Conclusions:

• BLM offers an effective and computationally efficient solution for cuffless BP estimation.
• Incremental BLM (IBLM) provides a scalable approach for continuous monitoring with reduced storage demands.
• The proposed BLM holds significant potential for personalized, real-time healthcare applications in BP monitoring.