Pressure ulcer evaluation using Machine Learning techniques

Abstract

Due to the rise of artificial intelligence and its new trends in machine learning models, several software products have been proposed to benefit the healthcare field. However, it was identified that medical personnel use different tools to evaluate pressure ulcers and diagnose their severity, which present a low degree of reliability. Consequently, we propose a software module with the use of different machine learning techniques, concluding with positive results in the level of diagnosis and reliability. In this article, we propose a convolutional neural networks model that allows to diagnose bedsores by means of captured images and to consult the severity of it in a fast way. For the validations of our proposal, different images of bedsores obtained from sources available on the Internet were used and the classifications of wounds diagnosed by a medical specialist were corroborated, being our proposed model the one in charge of revalidating these results.