Cuffless and Continuous Blood Pressure Estimation From PPG Signals Using Recurrent Neural Networks

Area of Science:

• Biomedical Engineering
• Medical Signal Processing
• Artificial Intelligence in Healthcare

Background:

• Continuous blood pressure monitoring is crucial for managing cardiovascular health.
• Traditional cuff-based methods are invasive and provide only intermittent readings.
• Photoplethysmography (PPG) signals offer a non-invasive alternative for physiological monitoring.

Purpose of the Study:

• To develop and validate a cuffless method for continuous blood pressure estimation.
• To leverage advanced recurrent neural network (RNN) models for improved accuracy.
• To identify optimal features from PPG signals for reliable blood pressure prediction.

Main Methods:

• Utilized Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) recurrent neural networks.
• Extracted and selected key features from PPG waveforms using machine learning techniques.
• Validated models on the Multiparameter Intelligent Monitoring in Intensive Care (MIMIC) database with diverse BP and PPG signals.

Main Results:

• The proposed RNN models demonstrated superior performance compared to traditional methods.
• Feature selection effectively reduced input dimensionality while maintaining accuracy.
• Achieved accuracy compliant with the Association for the Advancement of Medical Instrumentation (AAMI) standards.

Conclusions:

• Cuffless blood pressure estimation using PPG and advanced RNNs is feasible and accurate.
• This technology has the potential to enable continuous, non-invasive cardiovascular health monitoring.
• The findings support the clinical viability of the proposed method for widespread adoption.