Riscos de Privacidade em Dados de Saúde: Investigando Inferência de Atributo no DATASUS
Year: 2025Authors: Ramon G. Gonze, Igor W. Lemes, Jussara M. Almeida, Marcos A. Gonçalves, Mário S. Alvim
Venue: Simpósio Brasileiro de Cibersegurança (SBSeg)
DOI: Coming soon
Abstract
Statistical dissemination of health data is crucial for the formulation and monitoring of public policies and scientific research, but it presents important challenges regarding the privacy of data subjects. In this work, we formally and experimentally evaluate the risks of inferring sensitive attributes in the DATASUS outpatient procedure dataset, which contains microdata since 1994 to the present day on millions of citizens. We identified serious privacy risks - for example, in some cases it is possible to identify sensitive attributes with an accuracy higher than 90% in almost 30% of the records in the database. These results led to the question of whether the platform is compliant with the Lei Geral de Proteção de Dados (LGPD).