USING THE NAIVE BAYES CLASSIFIER TO GENERATE INFANT DEATH RISK ALERTS

Abstract

GISSA is an intelligent system for health decision making focused on children maternal care. Alerts are generated in this system that involve the five health domains: clinical-epidemiological, normative, administrative, knowledge management and shared knowledge. The system intends to contribute to the reduction of child mortality in Brazil. This paper presents LAIS, an intelligent mechanism that uses machine learning to generate child death risk alerts in GISSA. A data mining methodology is used to obtain a learning model able to calculate the probability of a newborn dying. The tests show that the Naive Bayes classifier is the most suitable algorithm for this purpose, presenting good results with a ROC curve of 92.1%. The work brings together the SIM and SINASC public databases for the training of classification algorithms, identifying relationships between birth and death data of children under one year of age. The spread subsample balancer was used, during the methodological process, which applies subsampling, improving model results.