Machine learning models outperform linear models in predicting energy expenditure.
The study compared different models to predict energy expenditure using data from accelerometers placed on the body. They found that for wrist-worn accelerometers, machine learning models were more accurate than linear models. However, for hip and thigh-worn accelerometers, linear models performed similarly to machine learning models. This suggests that for certain placements, simpler linear models can be just as effective for predicting energy expenditure.