Predictive Analytics for Monitoring and Classifying Physical Activity Intensity Using Smartphone Sensors

The study outlines a predictive analytics system based on smartphone sensors for monitoring and classifying physical activity levels. Usually, conventional wearable technology needs assistance with accuracy and comfort. However, as it makes use of smartphone sensors, the proposed system does not need any extra hardware. It collects data from accelerometers, gyroscopes, and magnetometers; it pre-processes the information to remove noise and extracts features associated with activity patterns. Classifying activity intensity is accomplished using machine learning (ML) techniques like Random Forests (RF), Support Vector Machines (SVM), and Neural Networks (NN). A great deal of testing and validation guarantee dependability. Real-time inference feedback gives users more incentive and engagement instantly. Over time flexibility is ensured by constant model iteration based on user input. Higher performance (0.92), precision (0.90), recall (0.94), and F1-score (0.92) than existing systems are shown by the results. A user satisfaction study indicates that all activities have less delay for activity classification, and usability, accuracy, and real-time feedback have all improved noticeably.