Wearable Sensors Show Promise for Remote OA Assessment

Evaluating seniors with knee osteoarthritis (OA) in their own homes, via wearable sensors that relayed data to clinicians' offices, appeared feasible in a small trial.

Among 20 patients tested, 18 were able to manage the sensors successfully, and the results correlated reasonably well with task performance during clinic visits, according to Deepak Kumar, PT, PhD, of Boston University, and colleagues.

Intraclass correlation coefficients, measuring the degree of agreement between at-home and in-clinic results with standard functional tests, ranged from 0.59 to 0.87, the researchers reported in Arthritis Care & Research. Test-retest reliability for the at-home data, when participants repeated the tasks after 20 minutes, was rated at 0.85 to 0.95, which Kumar's group called good to excellent.

The differences between at-home and clinic results were largely due to faster performance at the office visits, the group observed. But, they noted, "[d]ata collected at home may have greater ecological validity than those collected in the lab," insofar as patients' abilities in their normal environments are what really matter.

Participants mostly gave high marks to the sensor-based testing in a post-study survey in which they were asked to rate the experience on seven aspects of the testing using 5-point scales. These included how they managed the internet connections (for live video-based instruction) and sensor placement, comfort levels during testing, and whether they would want to use the technology in the future. Median scores were 5 on all but two ratings, and were 4 for those two ("overall experience" and "likelihood of participating" in additional at-home sessions).

The study was an initial analysis of data from a clinical trial whose stated aim was to determine whether wearable sensors could be used to evaluate outcomes in tests of exercise interventions in knee OA. (Pfizer and Eli Lilly helped fund the trial, suggesting an interest in the technology's potential in drug studies as well.) The full trial enrolled 60 patients; the current analysis focused on 20 of them and their general experience with the sensors and associated technology.

The at-home system consisted of three Opal inertial sensors, one worn on each foot and one strapped to the lower back. Participants connected with clinicians via Zoom who guided them on applying and activating the sensors and performing tests including walking two laps of a 7-meter path at a comfortable pace and performing five repetitions of standing from a seated position. The outcome in each was participants' speed in completing the tasks.

Mean participant age was 70.5 and their mean body mass index was 30.6. Self-rated pain in their most-affected knee averaged about 60 on a 100-point scale, and mean functional disability was 69, also out of 100. Women were 85% of the sample.

For two participants, the system failed to yield useable data. These patients had "challenges" following the Zoom instructions, Kumar and colleagues wrote, "resulting in battery depletion and lack of data collection in some sensors."

The researchers concluded that the technology is feasible, at least in the research setting, although it does require "rigorous methods" for guiding patients in its use, with both written and video instruction.

Limitations to the study included the small number of participants, the use of a particular type of sensor (such that generalizability to Fitbit-type devices isn't guaranteed), and lack of consistency in walking surfaces that might have influenced performance.

Comment