A Stanford study recently published in npj Digital Medicine suggests that the iPhone’s step measurement capabilities are sufficient enough to monitor and measure patients’ activity remotely. However, the device’s built-in walking distance measurement algorithm is inconsistent, and should not be used when gauging a patients’ activity or building a patient monitoring app for conditions such as peripheral artery disease (PAD).
Researchers found the iPhone’s CoreMotion Pedometer to underestimate users’ steps by a mean of just 7.2 percent (± 13.8 percent), and demonstrated a mean percent difference of 5.7 percent (± 20.5 percent) when compared to an ActiGraph GT9X Activity Monitor. On the other hand, the iPhone pedometer’s algorithm to estimate distance walked overestimated user’s travel by 43 percent (± 42 percent), which the researchers attributed to its inaccurate application of users’ stride length.
How it was done
To investigate whether the iPhone’s pedometer could be reliably used to measure the blood flow of PAD patients and their ability to walk, a team of Stanford researchers built an ResearchKit app that uses the iPhone’s pedometer to conduct six-minute walk tests (an objective measure of heart patients’ functional capacity). The team enrolled 114 participants with PAD to perform a monitored walk test while using the app and carrying the ActiGraph monitor. The app’s measurements during the test were then analyzed against manual step counts and the activity monitor’s readings.
What’s the history
Pedometers have been a feature of Apple’s consumer devices for quite some time, and some prior studies have contended that it and other smartphone’s tools are sometimes more accurate than dedicated fitness tracker wearables. While the present study was examining the feature for its viability as a medical screening tool — for which precision would be mandatory — the devices’ step counters have most often been used by the industry in various wellness and activity incentive programs.
“We demonstrated that the iPhone CMPedometer distance estimation algorithm has poor accuracy in patients with PAD, likely due to lack of correction for an individual’s stride length. Additionally, we showed that the iPhone CMPedometer step counting algorithm is highly concordant with the ActiGraph and reference standard. We therefore feel confident that steps measured by an iPhone app can be used as a metric for remote physical activity tracking in PAD patients.”