A study led by DCRI and Duke researchers was the first to comprehensively explore skin tone as a potential factor affecting accuracy of heart rate measurements from wearable devices.
A recent study published in Nature Partner Journals: Digital Medicine reveals new information about inaccuracies in wearable devices that measure heart rate.
Previous research had suggested that wearables did not function with the same accuracy across all skin tones; one study demonstrated that inaccurate measurements occurred up to 15 percent more frequently in participants with dark skin than those with light skin. The study, which was conducted by DCRI and Duke researchers, was the first to comprehensively explore skin tone as a potential factor affecting heart rate accuracy.
The study, which included 53 participants that represented the full range of skin tones on the Fitzpatrick skin tone scale, evaluated six wearable devices—four consumer-grade and two research-grade—as each participant completed a series of activities three times. An electrocardiogram was used as the reference standard.
The study found no statistically significant difference in accuracy across skin tones. In addition to skin tone, the study also investigated potential relationships between error in heart rate measurements and activity condition, type of wearable device, and wearable device category. Important findings from the study included:
- All three of the other factors investigated—aside from skin tone—had a significant correlation with heart rate measurement error.
- Absolute error in heart rate measurement was, on average, 30 percent higher during activity periods than at rest.
- Consumer-grade wearables were found to be more accurate than research-grade wearables at rest.
“With more than 121 million Americans expected to use wearable devices by 2021, it is important to recognize that the data gathered is not always reliable,” said the study’s lead author, DCRI’s Jessilyn Dunn, PhD, assistant professor of Biomedical Engineering and Biostatistics & Bioinformatics. “Accurate heart rate measurements are important at an individual level so that people do not exceed their maximum heart rate during exercise, which can trigger adverse cardiac events. Reliability of data is also critical for wider health care and research applications of wearable devices.”
“Despite these challenges, wearable devices have the potential to revolutionize health care, especially in traditionally underserved communities,” said first author Brinnae Bent, a graduate research associate in the BIG IDEAs Lab and Duke Forge predoctoral scholar. “This study takes a step toward validating the accuracy of wearables across the spectrum of skin tones.”
“As we enter the era of real world data and big data, this study illustrates the importance of validating and evaluating the quality of information conveyed through this data,” added the DCRI’s Benjamin Goldstein, PhD, who also contributed to the study.
Warren Kibbe, PhD, professor of Biostatistics & Bioinformatics at Duke University, was also a member of the study team. The DCRI’s Hwanhee Hong, PhD, contributed to the project’s statistical design.