Comparison and validation of pressure and acceleration time-domain waveform models of a smart insole for accurate step count in healthy people

Abstract

Several studies have shown good accuracies for step count based on pressure signals of smart insoles in people walking at different speeds. Although smart insoles are often equipped with pressure sensors and accelerometer, no study has focused on comparing the accuracy of step count separately based on pressure and acceleration signals in healthy people. The objectives of this study were to design a waveform model of accelerometer and pressure sensors, and then compare with commercially well-known step count devices and validate these models using manual step counter for step count. Eight healthy participants (age: 39.8±17.56 years old) wore a pair of smart insoles, a GaitUp, and a StepWatchTM and performed the six-minute walking test at walking speeds from 1.62 to 2.22 m/s. Four pressure and one acceleration waveform models were designed and used for the detection of 341 to 412 steps. Accuracies ranged from 99.80%±0.60% to 99.97%±1.38% for right side, and from 99.67%±0.63% to 99.90%±0.05% for left side with pressure waveform models. In addition, the acceleration waveform model provided accuracies of 99.87%±2.49% and 99.84%±4.77% for right and left sides respectively. Step count accuracies using the GaitUp were 99.51%±2.06% for right side, and 99.51%±4.32% for left side. Finally, the StepWatchTM yielded step count accuracies of 99.31%±15.95% and 98.52%±28.06% for right and left sides respectively. These results suggested the smart insole with pressure and acceleration waveform models as more accurate than the StepWatchTM and the GaitUp for step count.