Abstract: This study focuses on the sensor position in an insole system that aims to detect changes in physical conditions. To reduce the costs, the number of pressure sensors is limited to four. The system evaluates the changes in the load applied to each sensor on the insoles. Commercially available insoles are classified into S, M, and L sizes and cut to fit the shoe size. Consequently, sensors are not always attached at appropriate positions on the insole, and substantial variations are expected to occur because of misalignment. The output characteristics differ significantly depending on the toe sensor position. In particular, the toe length varies considerably among individuals, and the sensor position must be adjusted to suit each individual. The peak value of the sensor output and the steepest slope value at the subsequent decrease are promising feature values. The incorporation of machine learning into the output results, including other sensor positions, is expected to yield more accurate data.

Keywords: Pressure sensors, Insole, Health condition change, Arduino, Bluetooth, Classification, Machine learning.

___________________________________________________________________