Feature Extraction for Gait Analysis Using a Smart Insole.

Every day we take thousands of steps, most of the time without even thinking of it. However, the human locomotion is a complex process that involves different systems of the human body. For this reason, human gait can be affected by several pathologies with very different origins, ranging from musculoskeletal to neurological disorders. Therefore, it is important to be able to properly analyze human gait in order to detect and prevent gait related pathologies.

For example, in the case of runners, it is important to have a good running technique in order to prevent injuries. It is estimated that the 18% of the novice runners will suffer some kind of running related injury every 1000 hours of running. Moreover, in recent years there has been an increase on the amount of people who run couple of times a week.

However, in most of the cases, the people who start in running does not receive assistance from an expert. Thus, they tend to not follow a correct training schedule and they might have a bad running technique, which could cause serious injuries in the future. Also, in the past years there has been a development on the mobile Apps and wearables that allow the user to control better their training. Yet, the information that the mobile applications have is limited (GPS coordinates), so these applications are not enough to analyze the running technique.

On the other hand, over decades motion analysis studies have been carried out using motion capture systems or force platforms. Nevertheless, these technologies have one big limitation, they need to be used in specialized motion capture laboratories and have a high cost. Therefore, the analysis made by these systems cannot be done in real life environments and they can only be used to measure short periods of time.

In this work, we present a sensor insole with pressure and inertial sensors, that with the analysis of those signals it is capable of extracting a great number of characteristics related with the running performance. The sensor looks like a normal insole, and its placed inside the shoe like any normal insole. The data is transmitted over Bluetooth Low Energy (BLE) to any computer or Smartphone. Thus, this system allows the motion analysis to be done in any place and moment.

The developed algorithm was validated in an experiment with 6 healthy subject that walked and ran at different speeds in a treadmill. We obtained a high rate of activity detection (walking or running) and analyzed features that are related with the pronation and supination while running.