Sensor Fusion for Robust Heartbeat Detection during Driving.

Area of Science:

• Biomedical Engineering
• Health Monitoring Technologies
• Signal Processing

Background:

• Private spaces, such as vehicles, are underutilized for continuous health monitoring using biosignals.
• Existing methods often struggle with accuracy in noisy and dynamic environments like driving.

Purpose of the Study:

• To develop and evaluate a sensor fusion approach for robust heartbeat detection in a driving environment.
• To identify optimal sensor combinations for accurate heart rate measurement during rest and motion.

Main Methods:

• Utilized four biosensors: electrocardiography (ECG), ballistocardiography (BCG), photoplethysmography (PPG), and image-based PPG (iPPG).
• Employed convolutional neural networks (CNNs) for sensor fusion and heart rate measurement.
• Monitored twelve healthy volunteers under rest and driving conditions for 11 minutes each, using a 3-lead ECG as ground truth.

Main Results:

• Sensor fusion achieved high accuracy, with combined sensors reaching up to 97.15% (rest) and 93.08% (motion).
• Fusing all four signals yielded the highest scores: 97.24% (rest) and 94.38% (motion).
• Specific combinations like BCG + PPG and ECG + BCG showed superior performance in rest and motion, respectively.

Conclusions:

• Sensor fusion significantly enhances the robustness and accuracy of heartbeat measurement in dynamic environments.
• This technology supports continuous, unobtrusive health monitoring for drivers, facilitating early disease detection.
• Exploiting private spaces like vehicles for biosignal monitoring holds great potential for proactive healthcare.