La classification des Maladies via une Analyse Médicale Basée sur l’Apprentissage Profond

Abstract

One of the major challenges faced by doctors is making decisions regarding disease related to the patient’s condition. Our topic addresses one of the problems related to this, which is : can data be further used to improve the accuracy of medical decision-making ? To solve this problem, we propose a disease classification mechanism based on medical data and patientassociated symptoms, relying on deep learning algorithms. The proposed solution to the presented problem relies on building models for standard deep artificial networks by applying different optimization algorithms (SGD, ADAM) and applying PCA for dimensionality reduction. The results of comparing the performance of different models showed high efficiency in applying dimensionality reduction with the SGD optimization algorithm in terms of handling new data and the time required for training. We conclude from the results the effectiveness of deep learning in solving problems associated with classification and data exploitation.