The Recognition Method of Athlete Exercise Intensity Based on ECG and PCG

Area of Science:

• Sports Science
• Biomedical Engineering
• Artificial Intelligence

Background:

• Traditional athlete training relies on subjective experience and basic indicators like heart rate, often leading to inaccurate intensity assessment and potential injury.
• Existing methods lack precision, potentially causing physical harm or suboptimal training outcomes.

Purpose of the Study:

• To develop a precise method for recognizing human exercise intensity using ECG and PCG signals.
• To enable real-time monitoring and adjustment of athlete training plans based on accurate intensity recognition.

Main Methods:

• Fused ECG (electrocardiogram) and PCG (Phonocardiogram) signals into 2D images.
• Trained a Convolutional Neural Network (CNN) model on a labeled dataset of motion intensities.
• Utilized the trained CNN model to predict exercise intensity from ECG and PCG signals.

Main Results:

• Achieved a recognition accuracy of 95.68% for exercise intensity.
• Demonstrated that combined ECG and PCG signals provide richer feature information than heart rate alone.
• The developed model shows high accuracy and generalization ability for real-time monitoring.

Conclusions:

• The proposed ECG and PCG-based method, utilizing CNN, offers a highly accurate and reliable approach for monitoring athlete exercise intensity.
• This method allows for personalized training adjustments, enhancing training effectiveness and reducing injury risk.
• The findings suggest a significant advancement over traditional heart rate-based monitoring in sports science.