Psychological stress recognition from heart rate variability parameters based on field programmable gate arrays

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Signal Processing

Background:

• Psychological stress poses significant health risks, necessitating early detection methods.
• Electrocardiograph (ECG) signals offer a non-invasive source for monitoring physiological responses to stress.
• Existing stress detection methods may lack precision, portability, or real-time analysis capabilities.

Purpose of the Study:

• To design and develop an intelligent, portable device for early psychological stress recognition.
• To implement a novel Adaptive Weight Deep Forest (Aw-Deep Forest) model for enhanced stress analysis.
• To validate the device's efficacy in analyzing heart rate variability (HRV) from ECG data for stress assessment.

Main Methods:

• ECG signals acquired using the AD8232 module and processed on a low-power Field Programmable Gated Array (FPGA) PYNQ-Z2 board.
• Development and implementation of the Aw-Deep Forest algorithm for improved forest fitting quality.
• Analysis of heart rate variability (HRV) parameters derived from filtered and digitized ECG signals.

Main Results:

• The developed device demonstrates intelligence, precision, portability, fast response, and low power consumption.
• The Aw-Deep Forest model shows improved performance over standard Deep Forest models.
• The integrated system achieved a final accuracy of 81.39% for psychological stress recognition.

Conclusions:

• The proposed device effectively utilizes ECG signals and the Aw-Deep Forest model for accurate psychological stress detection.
• FPGA implementation enables real-time, low-power stress assessment.
• This technology holds promise for non-invasive, accessible mental health monitoring.