Comprehensive Health Tracking Through Machine Learning and Wearable Technology

Yusuf, Abusufyan; Jaber, Tareq Al; Gordon, Neil

Abstract

The accurate interpretation of data from wearable devices is paramount in advancing personalized healthcare and disease prevention. This study explores the application of machine learning techniques to improve the interpretation of health metrics from wearable technology, focusing on heart rate and activity prediction. The study conducts a device-wise comparison of data from popular devices, namely the Apple Watch and Fitbit, using both tree-based and boosting algorithms. The outcome of the experiment shows that the Random Forest model is a better predictor for heart rate, with the lowest error rate across devices and a prediction accuracy of 98% on the combined dataset. Conversely, the classification result for activity prediction showed that all models used have better accuracy with Fitbit data, and accuracy drops with Apple Watch data. The Random Forest achieves a consistent performance of 87% for accuracy and F1 score on the combined data. However, after cross-validated hyperparameter tuning, this result on the combined dataset is superseded by the boosted models, with both Gradient Boosting and XGBoost achieving the same level of performance (90%) across metrics.

Citation

Yusuf, A., Jaber, T. A., & Gordon, N. (online). Comprehensive Health Tracking Through Machine Learning and Wearable Technology. Journal of Data Science and Intelligent Systems, https://doi.org/10.47852/bonviewjdsis52023588

• Journal Article Type: Article
• Acceptance Date: Feb 19, 2025
• Online Publication Date: Mar 11, 2025
• Deposit Date: Mar 11, 2025
• Publicly Available Date: Mar 11, 2025
• Journal: Journal of Data Science and Intelligent Systems
• Print ISSN: 2972-3841
• Electronic ISSN: 2972-3841
• Publisher: Bon View Publishing
• Peer Reviewed: Peer Reviewed
• DOI: https://doi.org/10.47852/bonviewjdsis52023588
• Keywords: Wearable device technology; Health tracking; Machine learning; Activity prediction, data science
• Public URL: https://hull-repository.worktribe.com/output/5077459
• Publisher URL: https://ojs.bonviewpress.com/index.php/jdsis/index