October 29, 2026 | Erin Bluvas, bluvase@sc.edu

The data is already there. Today’s smartwatches, rings, monitors and other wearable devices are continuously collecting information about our health. But we only have access to the most basic metrics, like step count and heart rate. Some products take it a step further and attempt to provide sleep scores and other more complex measures of health, but it really only scratches the surface.

With support from a $3.2 million NIH grant, a team of researchers will focus their expertise on making the most of this data – applying combinations of existing and novel biostatistical methods to open an entirely new level of health insights that are more useful and accurate than what we’re used to. The project will be led by biostatistics assistant professor Rahul Ghosal.

“Wearable and sensor technologies now record physiological signals continuously as people go about their daily lives, including cardiac activity (ECG), brain activity (EEG), physical activity (PA), and continuous glucose monitoring (CGM),” Ghosal says. “These high-frequency streams capture the timing, variability, and interplay of physiological processes, and they can reveal subtle but clinically meaningful behavioral phenotypes. However, most current analyses discard this richness.”

By reducing the data to a few summary statistics, like total activity and amount of moderate-to-vigorous physical activity, the typical information reported from these devices falls far short of its potential. By applying and integrating advanced statistical methods, these products could provide much more meaningful signals of health and disease.

Working with University of Minnesota’s Erjia Cui, Ghosal and the team will use the five-year grant from the National Heart Lung and Blood Institute to unveil the full shape of a person’s activity distribution and temporal pattern across the 24-hour day. They will do this by developing flexible statistical methods that unite functional data analysis (allowing them to model the temporal trajectory) and distributional data analysis (capturing personalized distributional characteristics such as intensity, volume, variability and physiological reserves within a single modelling framework).

The researchers will also account for the limitations and complexities of today’s wearable device data, such as integrating multiple sources, and irregular or sparse measurements. Initially, they will focus on nationally representative samples of older adults. Their goal is to outline patterns of how individuals’ 24-hour activity patterns evolve over time longitudinally and discover clinically useful phenotypes of aging that predict cumulative and specific health risks, such as cardiovascular disease, cognitive impairment, and changes in functional independence among older adults.