Cloud-Based Predictive Modeling of Energy Expenditure from Wearable Data using Random Forests

Energy expenditure (EE) estimation using physiological signs is becoming more important as wearable devices proliferate. To estimating EE from wearable data, this research proposes a cloud-based predictive modeling strategy based on Random Forests. In the approach, it gathers a variety of physiological signals, as well as accelerometer and heart rate data, from several wearable devices. A Random Forest model, hosted on the cloud, is then given this pre-processed data. To test the method, it uses a large dataset including information on people's physical activity levels across a wide range of activities. When tested against more conventional approaches, the findings show that the proposed cloud-based prediction model outperforms them in terms of EE estimation accuracy. The use of cloud computing also makes real-time EE estimate application more accessible and scalable. Implications for individualized fitness monitoring and healthcare treatments stem from this paper contribution to the area of predictive modeling for wellness and health monitoring via wearable devices.