A hybrid hierarchical framework for free weight exercise recognition and intensity measurement with accelerometer and ECG data fusion

Accurate recognition and effective monitoring of physical activities (PA) in daily life is a goal of many healthcare fields. Existing PA recognition approaches are mostly designed for specific scenarios and often lack extensibility for application in other areas, thereby limiting their usefulness. In this paper, we present a hybrid hierarchical framework that successfully combines two of the main specific-sensor-based PA methods into an effective hybrid solution for general weight exercise applications. The fusion solution separates free weight and non-free weight activities and then further classifies free weight exercises, whilst measuring quantities of repetitions and sets, thus providing a measure of intensity. By fusing accelerometer and electrocardiogram (ECG) data, a One Class Support Vector Machine (OC-SVM) and a Hidden Markov Model (HMM) are applied for exercise recognition and we use semantic inference for determining the intensity of the exercise. The results are based on data from 10 healthy subjects (age: 30 ± 5; BMI: 25 ± 5.5 kg/m^2; body fat: 20.5 ± 5.4), which shows good accuracy in exercise recognition and intensity measurement. This framework can be extended to support additional types of PA recognition in complex applications.