Detecting Exercise-Induced Temperature Distribution Changes at the Knee Using a Wearable Array of Thermistors

The Heat Distribution Index (HDI) has been used to capture the abnormal variance in knee temperature distribution as a proxy for inflammation in osteoarthritis. HDI has been derived from thermal camera (TC) data, but its use has been limited to static conditions, as TCs are impractical for dynamic assessments. This study evaluates the feasibility of using a wearable array of thermistors to assess HDI changes in response to exercise, using TC data as reference. Ten healthy participants were enrolled, and knee skin temperature was measured pre-, post-exercise, and during a cooldown period of approximately 30 min. The thermistor-based system detected a significant HDI increase post-exercise ( $\Delta HDI=0.76^{\circ }\text {C}, p\lt 0.001$ ) consistent with the TC results ( $\mathrm {\Delta HDI=0.71^{\circ }C, p\lt 0.001)}$ , with no statistical difference detected between the two methods $\mathrm {(p\gt 0.05)}$ . During cooldown, both systems tracked recovery trends without detectable differences $\mathrm {(p\gt 0.05)}$ . The mean HDI error for the thermistor array was 0.20±0.36∘C. Simulations assessing performance and sources of error showed negligible bias in HDI estimates regardless of the number of thermistors, while variance decreased as thermistor count increased. A more uniform sensor distribution across the region of interest improved accuracy over clustering near its edges, while misplacement had minimal impact except for rare outliers. These findings support the feasibility of a wearable array of thermistors for assessing exercise-induced HDI changes. Improvements in sensor placement to maximize skin contact while allowing for perspiration may enhance reliability. This work lays the foundation for real-time thermal monitoring and may aid in developing wearable tools to monitor patients with knee osteoarthritis.