Classifying work rate from heart rate measurements using an adaptive neuro-fuzzy inference system

Area of Science:

• Exercise Physiology
• Computational Intelligence
• Biomedical Engineering

Background:

• Accurate work rate classification is crucial for exercise prescription and monitoring.
• Current methods using percent heart rate reserve (%HRR) have limitations in accuracy and inter-participant variability.
• Physiological and physical differences among individuals pose challenges for standardized work rate assessment.

Purpose of the Study:

• To develop and validate a novel Adaptive Neuro-Fuzzy Inference System (ANFIS) for classifying work rate.
• To compare the performance of the ANFIS classifier against established methods based on %HRR.
• To investigate the significance of heart rate monitoring (HR, HRmax, HRrest) and body weight in work rate classification.

Main Methods:

• Utilized field heart rate (HR) measurements from 28 participants during step and treadmill tests.
• Employed Adaptive Neuro-Fuzzy Inference Systems (ANFIS) to classify work rate into four categories: very light, light, moderate, and heavy.
• Measured heart rate and oxygen consumption (VO2); considered inter-participant variability.

Main Results:

• Heart rate monitoring (HR, HRmax, HRrest) and body weight were identified as significant variables for work rate classification.
• The ANFIS classifier demonstrated superior sensitivity (90.7%) and specificity (95.2%) compared to %HRR methods.
• ANFIS achieved a 29.6% greater classification accuracy on average, balancing sensitivity and specificity effectively.

Conclusions:

• The ANFIS classifier offers a more accurate and reliable method for assessing work rate compared to traditional approaches.
• The system effectively accounts for inter-participant variability, enhancing its practical application.
• The ANFIS classifier's ease of implementation and reliance on measurable variables suggest potential for widespread adoption by practitioners.