Key result
Beta-blocker treatment lowered average heart rate from 77.7 to 74 bpm and improved DL model scores in 2,914 wearable-monitored individuals.
Why the study?
The potential of wearable technology for assessing cardiovascular care quality remains untapped due to a lack of large datasets linking electronic health records with wearable metrics.
Does beta-blocker therapy reduce daily average heart rate as measured by longitudinal wearable data?
Does beta-blocker therapy reduce daily average heart rate as measured by longitudinal wearable data?
Longitudinal wearable data linked with electronic health records can effectively monitor physiological responses to beta-blocker therapy, offering a scalable approach for evaluating treatment efficacy.
Answers are grounded in this paper and cite their sources.
Supports wearable-EHR monitoring of beta-blocker response; leaves open prospective validation before clinical adoption.
Background Wearable technology presents unique opportunities for large-scale screening of cardiovascular disease, yet its potential for assessing care quality remains untapped. A major obstacle has been the lack of large datasets linking electronic health records (EHR) with wearable-generated metrics. The All of Us (AoU) Research Program addresses this gap by integrating longitudinal wearable data with EHR, providing a powerful platform to develop innovative methods for monitoring cardiovascular care quality. Purpose We demonstrate the utility of wearables for long-term monitoring of heart rate as an efficacy endpoint in individuals treated with beta-blockers in the AoU Research Program. Methods Participants with available wearable data and EHR in the AoU were included. Individuals with a prescription for a beta blocker were identified and daily average heart rates were compared before and after initiation of therapy. To further capture the heart rate variability patterns, a deep learning classifier was trained to predict the daily use of beta blockers from the minute-level heart rates after splitting the data at the participant level between the train and test sets. After alignment of the first date of medication use, changes in the heart rate and model-predicted scores were assessed in subgroups of sex and race as captured in the EHR. Results A total of 2,914 unique individuals with 1,044,053 days of collected wearable data were included (mean age 58.2 ± 15.4 years, 62.7% female, median follow-up 144 days/participant). Overall, 879 participants had a history of arrhythmia (30%), 826 had coronary artery disease (28%), and 394 had heart failure (13%). The deep learning model showed modest discrimination of daily use (AUROC = 0.7, AUPRC = 0.6) and improved discrimination of weekly use of beta-blockers (AUROC = 0.73, AUPRC = 0.64, Figure 1). Treatment with a beta-blocker was associated with a steep decrease in the heart rate (Pre-treatment 77.7±10.6 bpm to post-treatment 74±10.5 bpm, p-value<0.001) and increase in model-predicted scores (pre-treatment 0.48±0.16 to post-treatment 0.57±0.16, p-value < 0.001, Figure 2). Immediate changes in the heart rate and model-predicted scores were significant in all demographic subgroups, including males, females, non-Hispanic White, and non-Hispanic Black participants. Conclusion Our findings highlight the potential of wearable technology in evaluating the treatment response to beta-blockers, offering a novel, individualized, and scalable approach to evaluating their treatment efficacy. Deep learning models can identify nuanced patterns in high-dimensional wearable data to personalize clinical decisions. The EHR-linked wearable data presents new opportunities to evaluate cardiovascular care quality, uncover possible disparities in therapeutic response, and personalize treatment monitoring.
No takes yet. Share an insight, caveat, or question.
Nargesi et al. (2025) studied this question. Beta-blocker treatment lowered average heart rate from 77.7 to 74 bpm and improved DL model scores in 2,914 wearable-monitored individuals.
Synapse has enriched 10 related papers. Consider them for comparative context: