Using Kinect body joint detection system to predict energy expenditures during physical activities.

Area of Science:

• Biomechanics
• Human Movement Analysis
• Exercise Physiology

Background:

• Exergames offer engaging physical activity but accurately measuring energy expenditure remains a challenge.
• Traditional methods for energy expenditure assessment can be cumbersome or expensive.
• Kinect's motion-sensing capabilities present a potential low-cost solution for activity monitoring.

Purpose of the Study:

• To evaluate the feasibility of using Kinect body joint data for calculating energy expenditure during exergames.
• To develop and validate biomechanical models for estimating energy expenditure using Kinect technology.

Main Methods:

• Two Kinect-based biomechanical models, Kinetic Energy (KineticE) and Work Energy (WorkE), were developed.
• Regression models were created using data from 19 participants and validated on 5 others.
• Energy expenditure was estimated and compared against indirect calorimetry (QuarkAEE) and accelerometer data.

Main Results:

• The accelerometer showed the highest correlation (r=0.835) with QuarkAEE, followed by WorkE (r=0.805) and KineticE (r=0.466).
• Kinect-based models (AEEKinect and AEEWork) demonstrated significant agreement with QuarkAEE (r=0.641 and r=0.793, respectively).
• The developed models were comparable in accuracy to accelerometer-based predictions (r=0.682).

Conclusions:

• Kinect technology offers a viable and potentially accurate alternative for measuring energy expenditure during exergames.
• The proposed Kinect-based models are particularly effective for activities involving significant body movements.
• This approach could enhance the assessment of physical activity intensity in interactive exercise environments.