Portable sensor based dynamic estimation of human oxygen uptake via nonlinear multivariable modelling.

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Wearable Sensor Technology

Background:

• Noninvasive portable sensors offer convenience in biomedical applications.
• Accurate estimation of oxygen uptake (VO2) is crucial for assessing exercise intensity and energy expenditure.
• Existing methods may lack accuracy for dynamic or complex exercise scenarios.

Purpose of the Study:

• To develop and validate a multivariable Hammerstein model for estimating human oxygen uptake (VO2) during treadmill exercise.
• To leverage data from wireless heart rate sensors and triaxial accelerometers for improved VO2 estimation.
• To compare the performance of the proposed multivariable model against single-input, single-output models.

Main Methods:

• Utilized a multivariable Hammerstein model identification approach.
• Employed pseudo-random binary sequence (PRBS) exercise protocols to differentiate linear and nonlinear system dynamics.
• Applied Support Vector Machine (SVM) regression for nonlinear static modeling and Multivariable Autoregressive with Exogenous inputs (ARX) for dynamic modeling.

Main Results:

• The developed nonlinear multivariable model demonstrated superior VO2 estimation accuracy compared to single-input, single-output models.
• The Hammerstein model effectively captured the complex physiological responses during exercise.
• The model successfully integrated data from multiple portable sensors.

Conclusions:

• The established multivariable Hammerstein model provides a robust and accurate method for estimating oxygen uptake during treadmill exercise.
• This approach enhances the utility of noninvasive portable sensors for physiological monitoring.
• The model shows potential for dynamic energy expenditure estimation in outdoor exercise settings.