Exploration of Wearable Device Data in Predicting Clinical Outcomes
Institute or Center: National Institute of Biomedical Imaging and Bioengineering (NIBIB)
Project: Exploration of Wearable Device Data in Predicting Clinical Outcomes
- Cloud computing, machine learning (ML), and deep learning (DL)
- Programming in commonly used languages in ML/DL and data science, such as Python
- Time series data analysis (experience with wearable sensor data analysis preferred)
- Experience with harmonizing and aggregating sensor data across different device types
- Experience with developing models, algorithms, and software for clinical decision support
About the position: The NIBIB seeks a data scientist to aggregate and analyze wearable device data from different sources to draw meaningful scientific insights, such as detection, risk stratification, or prediction of disease. The scholar will leverage large volumes of data and accompanying machine learning models housed in an NIH data hub to address multiple interrelated research questions:
- How FAIR are the data and models in the hub?
- How accurate, reproducible, and generalizable are the findings reported by the groups who generated the data?
- To what extent can wearable data from different devices and studies be aggregated? To what extent can predictive models from one study be run on wearable data from another study?
- Can standards for data collection and harmonization be established to enable predictive models from aggregated data of different devices and studies to improve public health?
- How can reliable data and computational technologies, such as machine learning, help to improve the detection, monitoring, prediction, prevention, or management of diseases?
About the work: Wearable device data from different manufacturers comes in different forms, and the challenges to create predictive models at population levels remain poorly understood. Through this project, the scholar will improve our understanding of how wearable device data can be aggregated and analyzed, and to establish standards for data collection and harmonization that can be shared with the scientific community and the wearable device industry.
Datasets included: Early in the COVID pandemic, NIBIB and NCI funded several projects to explore how wearable device data can be used to predict COVID-related outcomes. Using different devices, the studies collected a variety of data including features of heart rate and respiratory rate, combined with symptom surveys and clinical outcomes. Machine learning models were developed to make predictions from these data. The data and models reside in a hub at the MIT Lincoln Laboratory.
Why this project matters: This project will have an immediate impact on how NIH approaches wearable device research – in areas related to data interoperability, standards, and FAIRness. The scholar will also help develop and validate prototype models and software based on aggregated data to support clinical decisions for healthcare professionals, caregivers, patients, or the general public, which may be deployed in real-world applications.
Work Location: Bethesda, MD with remote and telework options
Work environment: The scholar will be senior special project leader within NIBIB and interact closely with experts in digital health, point-of-care, and data/computer sciences in the DHIT division. The scholar will report directly to Dr. Behrouz Shabestari, who directs DHIT. Program staff in the NCI Center for Strategic Scientific Initiatives (CSSI) will remain engaged throughout the project.
To apply to this or other DATA Scholar positions, please see instructions here: datascience.nih.gov/data-scholars-2022.
This page last reviewed on April 7, 2022