The participants were asked to perform three exercise tests while wearing the device. First, they had to stand from a sitting position 10 times. Second, they were asked to extend the leg from the knee while sitting down (flexion-extension) 10 times. Third, and finally, they were asked to perform a one leg stand on the right leg twice.
For the acoustic data analysis, data from the standing phase of the sit-to-stand test and the extension phase of the flexion-extension tests were used. Mr. Tuilpin explained that the acoustic signals underwent processing to segment and filter them into candidate locations. The average sound patterns seen in the candidate locations were then analyzed, then a consistency analysis was undertaken. With this approach, inconsistent patterns of knee crepitation could be captured, Mr. Tiulpin explained.
Kinematic signals received from the movement sensors were used to determine the degree of knee instability. Higher signal magnitudes could potentially indicate stability problems, which can be quantified using signal power, Mr. Tiulpin’s slides stated.
A variety of statistical calculations were made to see how well the device might predict OA changes, and a model combining body mass index and age had an area under the curve of 84%, which suggested that it might be possible to improve OA detection with the addition of the device versus BMI and age alone.