0
0
West Virginia University Study Suggests the Need for Improved Algorithms in Wearable Technology
In an era where smartwatches and fitness rings have become everyday tools for monitoring health, a recent study by researchers at West Virginia University (WVU) has raised questions about the accuracy of these devices. Specifically, the study found that the heart rate variability (HRV) data reported by wearables differs from those obtained in clinical settings, potentially leading to biased health assessments.
“Heart rate variability has been a key non-invasive measure of the autonomic nervous system for nearly a century, providing insights into overall mortality, cardiovascular health, and stress,” said Matt Tenan, program director for Human Performance Research and Data Science at the WVU Rockefeller Neuroscience Institute. “While consumer wearables are now offering HRV data that was once only available in hospitals or labs, they’re measuring it differently.”
The research team highlighted that wearables use a technology known as photoplethysmography (PPG), which involves shining light into the skin to measure blood flow just below the surface. In contrast, clinical settings utilize electrocardiography (ECG or EKG) machines, which record the heart’s electrical activity via electrodes placed on the body.
“You’re looking at two things: one is blood flow, and the other is the electrical signal of the heart,” Tenan explained.
HRV is of significant interest to clinicians and medical scientists because it serves as a biomarker for a patient’s overall systemic health. Although users of wearable devices might not focus directly on HRV, it influences other wellness metrics provided by the device, such as readiness or sleep scores.
“These scores are often used to gauge overall health and fitness improvement. However, if a key measure like HRV is biased, the accuracy of these scores comes into question,” Tenan added.
To assess the validity of HRV measurements from wearables, Tenan and his colleagues conducted a simulation analysis using data from prior research. Their findings, published in the journal Sports Medicine, revealed that the HRV data from wearables, which measure the pulse, do not align with HRV data from ECG, which measures the heart’s electrical signals.
“There’s a lot happening between the heart’s beat and when the blood reaches the arm or wrist—a process known as pulse arrival time,” Tenan said. “Our simulation showed that what consumer wearables measure is different from what an ECG would capture. This doesn’t mean the wearables are useless, but it does mean they aren’t the same.”
The study also found disparities among different wearable brands in how they calculate HRV. The researchers discovered that Apple’s method, known as the standard deviation of normal-to-normal intervals (SDNN), is the most accurate. In contrast, many other brands use a method called root mean square standard deviation (RMSSD), which, according to Tenan, has a broader margin of error.
“I don’t see any reason why wearable companies should continue using RMSSD when SDNN offers greater accuracy,” Tenan stated.
The study’s findings could have significant implications for the development of machine learning algorithms for health monitoring in wearables. Tenan also expressed interest in conducting future research that would be more clinically focused, involving direct patient work.
This research underscores the importance of continuous refinement in wearable technology, particularly as these devices become increasingly integrated into daily life and personal health management.
More Information: Hayden G. Dewig et al, “Are Wearable Photoplethysmogram-Based Heart Rate Variability Measures Equivalent to Electrocardiogram? A Simulation Study,” Sports Medicine (2024). DOI: 10.1007/s40279-024-02066-5.
