High-Resolution Wearable Device Data Used to Predict Cardiometabolic Disease Risk
Researchers developed a framework to extract high-resolution numerical phenotypes from wearable devices and use them to predict cardiometabolic disease risk.
According to a study published in the Journal of Internet Medical Research suggests.
Consumer wearable devices such as smartwatches and fitness trackers record heart rate, step count and other health data under normal daily conditions. Recent research has also demonstrated that summary statistics from these wearable devices have potential uses for longitudinal health and disease surveillance.
“Unlike clean data from controlled experimental settings, real-world portable recordings tend to be jagged, contain missing chunks, lack clean context annotations, and have varying lengths,” the authors of the study. “As such, analyzes based on the naïve application of general purpose time series feature extraction methods may not have ecological validity.”
For these reasons, the authors hypothesized that higher-resolution physiological dynamics and phenotypes recorded by wearable devices might be applicable to modifiable and inherent markers of cardiometabolic disease risk.
To observe this, the authors used a framework to extract high-resolution phenotype data from wearable devices and applied it to a multimodal dataset, using machine learning to model nonlinear relationships and model comparisons to assess the predictive value of high-resolution phenotypes.
They found that these high-resolution physiological characteristics had higher predictive value compared to typical reference values for clinical markers of cardiometabolic disease risk.
Compared to benchmarks, models that performed best using high-resolution features had a 17.9% higher Brier score when based on age and gender, and a 7% higher score .36% when based on resting heart rate.
Heart rates in different states of activity also contained different types of information, the authors found.
“Heart rate dynamics in sedentary states are the most predictive of lipid abnormalities and obesity, while patterns in active states are the most predictive of blood pressure abnormalities,” they said. “Additionally, compared to standard measurements, higher resolution patterns in wearable heart rate recordings are better able to represent subtle physiological dynamics related to genomic risk of cardiometabolic disease.”
According to the authors, the higher-resolution phenotypes resulted in an improvement in Brier scores between 11.9% and 22.0% for predicting genomic risk. Additionally, case studies have shown that there are links between high-resolution phenotypes and clinical events.
Based on these results, the authors highlighted 2 potential applications of the developed framework.
First, the study revealed novel relationships between high-resolution heart rate dynamics and cardiometabolic disease risk.
“These results highlight the added value of assessing physiology in free-living activity states (beyond controlled clinical settings) for monitoring and managing disease risk,” the authors said. .
Second, the results offer a new perspective on the links between data collected by wearables and genetic predispositions in cardiometabolic diseases.
“As these associations did not appear to depend on the presence or absence of overt clinical risk markers, we postulate that high-resolution wearable device phenotypes may capture subtle subclinical physiological changes resulting from latent predispositions to disease,” they concluded.
Zhou W, Chan YE, Foo CS, et al. High-resolution digital phenotypes of consumer wearables and their applications in machine learning of cardiometabolic risk markers: a cohort study. J Med Internet Res. Published online July 29, 2022. doi:10.2196/34669