Use of infrared sensors for estimation of energy expenditure by elderly people living alone at home.

Tri-axial accelerometers (triax) can estimate energy expenditure with a correlation of approximately 0.9. We explored the idea of developing an alternative non-contact method to estimate energy expenditure. In this paper, we present a statistical analysis of infrared sensor data and tri-axial accelerometer data to investigate a relationship between the infrared sensor outputs and energy expenditure as estimated using a triaxial accelerometer. Parametric regression of triax data and sensor data did not show any global fitting parameters. Non parametric regression method of locally weighted scatter plot smoothing (lowess) using tri-cube kernel weights gives an underlying trend of energy estimate. As it is a non-parametric method, it does not give us any global parameters to estimate measure of energy expenditure. But it gives us a good visual assessment of relationship between motion sensor data and accelerometer data.