Evaluación del rendimiento de sistemas inteligentes basados en aprendizaje automático y aprendizaje profundo para la detección del asma en una señal audible de tos

Abstract

Artificial intelligence in recent decades has made a breakthrough, it is used in various fields, one of them is medicine, which allows early detection of diseases. Cough is a sonorous sound that originates from several factors in which asthma is included. Until now, the analysis of cough as the beginning of a serious disease has a limitation, since it is detected with old and uncomfortable measurement tools. For the users. With respect to artificial neurons, they are similar to the brain neurons of people, they interconnect with each other to carry out different processes, they receive information from the outside or in turn from other neurons, they have the ability to learn and, based on this, they carry out varieties of machine learning applications such as image classification, natural language processing, and time series prediction. In the following degree work, we seek to analyze the different neurons such as the base neuron that belongs to automatic learning and the convolutional neuron of deep learning, it will be determined which of the two neurons has the best performance for the early detection of cough with asthma. To carry out this procedure, we have the audible cough audios from the database provided by the University of Cambridge, a conditioning of the cough signal is carried out and it is applied to the different neurons.