Estimating energy expenditure from accelerometer data in healthy adults and patients with type 2 diabetes

The aim of this study was to develop specific prediction equations based on acceleration data measured at three body sites for estimating energy expenditure (EE) during static and active conditions in middle-aged and older adults with and without type 2 diabetes (T2D).

Forty patients with T2D (age: 40-74 yr, body mass index (BMI): 21-29.4 kg·m^-2) and healthy participants (age: 47-79 yr, BMI: 20.2-29.8 kg·m^-2) completed trials in both static conditions and treadmill walking. For all trials, gas exchange was monitored using indirect calorimetry and vector magnitude was calculated from acceleration data measured using inertial measurement units placed to the participant's center of mass (CM), hip and ankle. Stepwise multiple regression analyses were conducted to select relevant variables to include in the three EE prediction equations, and three Monte Carlo cross-validation procedures were used to evaluate each separate equation.

Vector magnitude (p < 0.0001) and personal data (gender, diabetes status and BMI; p < 0.0001) were used to develop three linear prediction equations to estimate EE during static conditions and walking. Cross-validation revealed similar robust coefficients of determination (R²: 0.81 to 0.85) and small bias (mean bias: 0.008 to -0.005 kcal·min^-1) for all three equations. However, the equation based on CM acceleration exhibited the lowest root mean square error (0.60 kcal·min^-1 vs. 0.65 and 0.69 kcal·min^-1 for the hip and ankle equations, respectively; p < 0.001).

The three equations based on acceleration data and participant characteristics accurately estimated EE during sedentary conditions and walking in middle-aged and older adults, with or without diabetes.