Aplicação da mineração de dados em bancos da segurança e saúde pública em acidentes de transporte

• Ano de defesa: 2009
• Autor(a) principal: Galvao, Noemi Dreyer [UNIFESP]
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Paulo (UNIFESP)
• Palavras-chave em Português: Informática em Saúde Pública; Acidentes de trabalho; Saúde Pública; Mineração; Data mining; Traffic accidents
• Link de acesso: http://repositorio.unifesp.br/handle/11600/8955

Introduction: With the scientific and technological evolution a large number of data have been collected and stored. In order to investigate such databases, the health informatics uses the Knowledge Discovery in Databases (KDD) methodologies, that is, knowledge discovery of the databases. Data Mining, one of the phases of KDD permits the investigation of data in search for standards, often not visible just by simple observation. Aim: To identify, through the use of the data mining technology, rules about traffic accidents making use of data of the Justice Secretariat and Public Security, of the Unified Health System municipal of Cuiabá (morbidity and mortality). Method: An exploratory, retrospective, observational, cross-sectional study of the databases of security and public health of the municipality of Cuiabá-MT, in 2006 was used. The three banks were related using the probabilistic method, through the free software RecLink. A hundred and thirty-nine (139) true pairs of road accident victims were obtained. In this related bank the mining data technology was applied, using the APriori algorithm, the software used was WEKA, also of a free domain. Results: A preliminary analysis in the pre-processing phase of the WEKA tool, showed that of the 139 victims of accidents, 80,6% were male, between 20-29 years of age (41,7%). Most of the victims were drivers (35,3%), the means of transportation used by the victim was the motorcycle (33,1%). Collision was the main cause of accident (51,8%) verified by the analysis. Most of the victims received medical assistance (87,1%), and the Municipal Emergency Hospital of Cuiabá (HPSMCBA) received most of the victims of this set of data (36,7%); in average each victim remained in hospital for five days. With the application of the APriori algorithm, ten best rules were created, six of them, indicated a useful and comprehensible knowledge to characterize the victims of accidents in Cuiabá. Conclusion: Based on these results, teaching and prevention programs can be established and so, it is worth considering the data mining technology as a powerful tool in the analysis of secondary data, helping the decision-making process with the extraction of useful knowledge of databases originated from the health information systems and public security.