Protocols Analyze Big Data From Small Devices

Cedars-Sinai investigators are working on a solution. In a recent study, they developed protocols for determining when a cardiologist review of abnormal heartbeat readings is warranted, potentially allowing health problems to be identified and treated.

Published in the journal Applied Clinical Informatics, the research focused on heart-rate data gleaned from devices such as the Apple Watch.

Since 2015, Cedars-Sinai has enabled patients to sync devices to upload data to its proprietary electronic health-record system app, MyCS-Link, and link it to their healthcare records. Raymond Duncan, MD, medical director of technology and architecture in Enterprise Information Services (EIS) at Cedars-Sinai, led that effort.

This approach expanded patient access to medical charts previously limited to physicians, a move spearheaded by Darren Dworkin, EIS senior vice president and chief information officer. Over the years, the data have expanded in volume and scope.

"Initially, step counts were reported, but later thousands of heart rates per patient per week were logged using smart devices," said Joshua Pevnick, MD, co-director of the Division of Informatics and associate professor of Medicine and Biomedical Sciences. "Many of these heart rates proved to be abnormally low or high." Pevnick was principal investigator for the study.

As of July 3, 2019, 7,128 patients had synced their devices. The Apple Watch now typically reports one heart-rate reading every 10 minutes. As data uploads outpaced the resources to screen them, concern rose about the surge of unreviewed data and the abnormal heart rates within it.

In response, Pevnick and the team designed their study to learn more about heart-rate datasets and how and when to alert a patient's doctor about out-of-range readings while also ensuring patient privacy. Physician informaticists met monthly to review abnormally low and high heart rates uploaded to MyCS-Link.

Once patients with potentially concerning heart rates were identified, the scientists conferred with clinical informaticists, including cardiologist Yaron Elad, MD, assistant professor of Biomedical Sciences and Medicine, to discuss each case. In a narrower subset of cases in which scientists felt it was warranted, Elad contacted patients’ physicians to alert them of findings.

The research team analyzed the data in a deidentified manner that protected patient identity, linking the heart rates only to patient age and sex.

After consultation with managing physicians and careful review of all datasets the scientists were able to develop:

• A heart-rate visualization dashboard to identify concerning heart rates.
• Experience regarding which combinations of heart rates and electronic health record data were most clinically worrisome.
• A protocol in which only concerning heart rates would trigger a cardiologist review that revealed protected health information.
• An overall framework for addressing other patient-initiated data.

"Our proposed protocol could help guide future efforts, especially for cases with large amounts of data and where abnormal values may represent concerning but treatable conditions," Pevnick said. The system also possibly could extend to other types of data, such as new symptoms suggesting suicidal ideation or chest pain, he added.

Funding: Research reported in this publication was supported by a research award from Cedars-Sinai Precision Health and supported by the National Institute on Aging of the National Institutes of Health under award number K23AG04918.