Uma análise exploratória de dados e o uso de aprendizado de máquina para classificação de doenças cardiovasculares

Abstract

The purpose of this project is to show a study based on data science for the elaboration of an analysis regarding cardiovascular diseases (CVD). In addition, to identify causes that may influence an individual to acquire this type of disease. Based on patient data and predictive analysis, to identify the probability of future results of people with CVD, in order to help the health system to obtain a better prognosis regarding their patients. Using for that, previous data to perform a study, through artificial intelligence, to determine whether there is CVD in the patient. In particular, the work provides an exploratory data analysis (EDA) seeking to find correlations between the data studied and CVD, as well as the use of artificial intelligence (AI) with five classification algorithms for predicting cases of cardiovascular disease. The algorithms predict the risk of acquiring a cardiovascular disease based on previous information from a database collected from patients. The dataset used was obtained from the Kaggle repository, found from the IEEEDataPort platform, in which, based on the study, blood pressure, cholesterol, age and BMI were found to have a higher correlation between the risk of obtaining a cardiovascular disease. The results on the dataset using the machine learning technique obtained a better result for the Random Florest method with an accuracy of 80%, F1-score of 82% and 78% for the possibility of not having or having a cardiovascular disease, respectively, using cross-validation with kfold equal to $5$.