| Title: | Causal discovery of digital biomarkers to predict long-term mental and physical health |
| Lead Institution: | ETH Zurich |
| Principal investigator: |
Application 96553
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
We aim to derive new biomarkers that relate to an individual's physical as well as mental health. We will assess how the derived patterns predict responses to illness (in particular Covid19) in the future and mortality risks. We will leverage our current research on modeling sleep quality and fatigue based on input from wrist-worn wearable sensors, and we will extend our own smaller studies to a population level using the UK Biobank. As previous work as well as our own research has identified, (physical and mental) health and well-being are highly subjective. One-week activity recordings allow to paint a much more detailed picture of individuals and their routine in unsupervised settings. We aim to derive new insights based on novel feature extraction methods adapted from our previous research. Besides deriving new biomarkers for health, we also aim to advance existing statistical methodology that currently struggles to detect effects on datasets much smaller than the UK Biobank. We thus hope that the derived biomarkers contribute to a better understanding of highly-subjective aspects of physical and mental health, which we hope will advance personalized health care in the long run. We hope derived biomarkers will also prove useful for other researchers during future projects using the UK Biobank. We further aim to generate new statistical methodology that enables the discovery and analysis of wearable sensor datasets much smaller than the UK Biobank. We estimate that we will answer our first research question regarding physical health (mortality prediction, hospital admission, time until recovery) within 12 to 18 months. Answering the following research questions will take significantly less time since data processing and feature extraction will be mostly completed. We estimate that answering research questions regarding mental health and Covid19 response will take around 6 months each. To evaluate causal inference methodology on small longitudinal we estimate another 6 to 12 months, depending on whether we find promising areas for improvement. In total, we thus estimate the project to be finished within 3 years.1 Publication
| Pub ID | Title | Author(s) | Year | Journal |
|---|---|---|---|---|
| 18902 | Beyond the hours slept: inconsistent sleep routines threaten mental health in 100,000 UK Biobank participants | Max Moebus (+1) | 2025 | BMC Public Health |
Enabling scientific discoveries that improve human health