Adapting Young Adults' In-Shoe Motion Sensor Gait Models for Knee Evaluation in Older Adults: A Study on...

The human knee joint is crucial for mobility, especially in older adults who are susceptible to conditions like osteoarthritis (OA). Traditionally, assessing knee health requires complex gait analysis in clinical settings, which limits opportunities for convenient and continuous monitoring. This study leverages advancements in wearable technology to explore the adaptation of models based on in-shoe motion sensors (IMS), initially trained on young adults, for evaluating knee function in older populations, both healthy and with OA. Data were collected from 44 older OA patients, presenting various levels of severity, and 20 healthy older adults, with a focus on key knee indicators: knee angle measures (S1 to S3), temporal gait parameters (S4 and S5), and knee angular jerk cost metrics (S6 to S8). The models effectively identified trends and differences across these indicators between the healthy group and the OA group. Notably, in indicators S1, S2, S3, S7, and S8, the models exhibited a large effect size in correlation with true values. These findings suggest that gait models derived from younger, healthy individuals are possible to be robustly adapted for non-invasive, everyday monitoring of knee health in older adults, offering valuable insights for the early detection and management of knee impairments. However, limitations such as fixed biases due to differences in measurement systems and sensor placement inaccuracies were identified. Future research will aim to enhance model precision by addressing these limitations through domain adaptation techniques and improved sensor calibration.