A fitness training optimization system based on heart rate prediction under different activities

Area of Science:

• Sports Science
• Biomedical Engineering
• Artificial Intelligence

Background:

• Heart rate monitoring is crucial for gauging exercise intensity and optimizing training.
• The five heart rate zone theory, based on maximal heart rate percentages, is widely used by athletes and individuals.
• Accurate prediction of heart rate during various activities is essential for effective exercise prescription.

Purpose of the Study:

• To propose an optimized training system leveraging dynamic heart rate prediction.
• To enable personalized exercise adjustments for achieving specific training intensities.
• To enhance the effectiveness of physical activities like walking, running, and rope jumping.

Main Methods:

• Development of a system utilizing four Long Short-Term Memory (LSTM) neural networks.
• One LSTM network dedicated to human activity recognition (HAR).
• Three LSTM networks specifically designed for predicting heart rate during different exercises.

Main Results:

• The system accurately predicts heart rate across walking, running, and rope jumping activities.
• Dynamic heart rate prediction allows for real-time feedback and exercise modification.
• The proposed system facilitates adherence to predetermined training heart rate zones.

Conclusions:

• The LSTM-based dynamic heart rate prediction system offers a novel approach to exercise optimization.
• This technology can empower individuals and athletes to train more effectively and safely.
• Personalized exercise guidance based on predicted heart rate enhances training outcomes.