Development of Plantar Pressure and SEMG-based Artificial Intelligence System for Detection of Plantar Ulceration

摘要

Diabetic Peripheral Neuropathy (DPN) is one of the main complications that affect the neural function associated with diabetes, and is also the cause for the development of plantar ulceration due to the loss of protective sensation. In order to assess and eventually prevent plantar ulceration, DPN must be detected. The existing clinical solutions in detecting DPN are used for static responses only, and the dynamic response must also be considered. These dynamic responses are measured using plantar pressure sensors and surface electromyography (SEMG). The plantar pressure sensors locate potential location for plantar ulceration, while the SEMG detects DPN by inspecting the muscle movements in the lower limb. Support Vector Machine (SVM), which aims at minimizing the true error rate, was used in classifying the muscle response gathered from the SEMG in diagnosing the volunteers as normal (N), diabetic mellitus (DM) and diabetic with neuropathy (DN), with an accuracy of 75.86%. Linear regression, which can mathematically distinguish variables, was used in diagnosing the plantar region of the volunteers where ulceration has a high possibility of occurring and is identified as the Metatarsal 1 region.