| Title: | Development of disease incidence and exacerbation prediction models from wearable physical activity tracker in healthy adults: Machine Learning Approach |
| Lead Institution: | ETH Zurich |
| Principal investigator: |
Application 102250
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
About
The aim of this research is to develop novel machine-learning models to predict incidence, hospitalization, and mortality due to chronic diseases in healthy adults. We will utilize the 24-hour physical activity data collected using wearable activity trackers. Chronic disease is a significant global health issue, which contributes to more than 50% of deaths worldwide. Physical activity is an important determinant of healthy aging, and substantial evidence has shown its protective effect against chronic disease and mortality. The pervasive use of wearables and smartphones enables objective monitoring of movement behaviors over a long time period. Passive behavioral data can be combined with a cutting-edge machine learning (ML) approach to improve clinical predictive accuracy and performance with faster processing speed. Past research, however, has typically assessed specific populations suffering from prevalent diseases, such as cardiovascular disease, cancers, diabetes, or obesity, with less attention on healthy/at-risk populations. Existing studies also have used simple, aggregate measures of physical activity such as step counts or total activity bouts. These approaches do not contextualize a comprehensive profile of physical activity over 24 hours and ignore complex dimensions of human activity. Furthermore, a lack of rigorous testing in an independent cohort hinders the generalizability of existing findings. To address these limitations, we propose a study to develop ML-based novel disease and exacerbation prediction models by leveraging 24-hour accelerometer data in a large cohort of healthy adults from the UK Biobank. We estimate the project will last 24-36 months from the initiation of the project. Our findings will guide preventive efforts for chronic disease in healthy individuals, particularly "seemingly" healthy adults. Furthermore, our work will provide the potential of utilizing digital devices in tailoring interventions at a personal level to lower the disease burden and improve disease prognosis.1 Publication
| Pub ID | Title | Author(s) | Year | Journal |
|---|---|---|---|---|
| 17614 | From wrist data to lifespan: elucidating inflammation-driven biological aging via activity rhythms captured by wearable devices | Jinjoo Shim (+5) | 2026 | npj Aging |
Enabling scientific discoveries that improve human health