DEEP LEARNING FOR HEALTHCARE MANAGEMENT AND DIAGNOSIS

Abstract

Deep Learning is a sub-area of Machine Learning, which deals with the recognition, processing, interpretation and classification of images, text, speech, etc. It has been investigated and applied to the identification of faces in social networks, texts in manuscripts, oral communication for human-computer interaction and health. This article aims to present applications of Deep Learning in healthcare. To do this, a research for works using the following descriptors was performed: "Deep Learning", "Health", "Management", "Diagnostics", "Network", "Deep Learning". 14 papers were identified, highlighting especially the areas of computer science, health. As a result, it was identified that most studies suggest the use of Deep Learning for clinical diagnosis, due to its excellent accuracy in the interpretation and identification of patterns of disease in medical imaging. This work also indicates that the implementation of  Deep Learning with back-propagation and CNN (convolutional neural networks) algorithms  in classification tasks, while dealing with nonlinear problems, can achieve superior accuracy, even outperforming human agents. This conclusion is especially interesting for the medical field, which conducts the analysis and diagnostics based mainly on images. The potential of this technique not only assists in medical decisions and the accuracy of the diagnosis, but also assists the medical specialist to suggest treatment measures, consequently to speed the process and make the care of a medical specialist, known as being of high cost and difficult to access, viable to all people.