Mathematical model to predict the probability of quilombolas developing metabolic syndrome with a health care flowchart

Abstract

To develop a mathematical model using Machine learning algorithms to predict the probability of quilombolas developing metabolic syndrome, as well as to propose a health care flowchart. Cross-sectional study using artificial intelligence. Having or not metabolic syndrome was adopted as the dependent variable. Bivariate analysis was performed comparing independent variables, anthropometric and biochemical indicators in relation to the presence of metabolic syndrome and categorical variables that were evaluated using the chi-square test (p <0.05). The Analysis of Variance or Kruskal-Wallis test was used according to the normality trend evaluated by the Shapiro-Wilk test. The machine learning data analysis tool was used, through the Decision Tree method. The decision tree for predicting metabolic syndrome in quilombolas was 75% accurate, generating a graph in relation to the decision process illustrated by means of a flowchart to guide decision-making in relation to health and prevention of metabolic syndrome. The predictive model allowed identifying the specific cutoff points of the most important anthropometric indicators to be measured in the first health care of the quilombolas. The accuracy of the predictive model allows the application of the flowchart in other quilombola communities, presenting itself as a technological tool that facilitates decision-making in health.