14...2026
10:00 Doctoral defense Room 85 of IC2
Topic on
An Open Health Data Infrastructure for Interoperability across Health Information Systems based on the Brazilian Ecosystem
Student
Marcia Jacobina Andrade da Silva Martins
Advisor / Teacher
Claudia Maria Bauzer Medeiros
Brief summary
Health Information Systems (HIS) depend on the integration of multiple data sources, and interoperability is still considered a major challenge in this area. This challenge is mainly related to the heterogeneity of data and systems and the fact that many curated biomedical resources are developed primarily in English. In the Brazilian context, official data sources for SIS include resources provided by ANVISA, the National Health Data Network, and official ICD-10 catalogs. In particular, information about medications is primarily available through package inserts published by ANVISA. However, this information is mostly provided in an unstructured format and presents variations in nomenclature and organization that hinder its automatic interpretation and its large-scale use. In fact, the mere availability of official health data does not guarantee its effective use in interoperability between health information systems. This thesis addresses the following research problem: how to transform unstructured textual data, available in Portuguese and originating from official Brazilian sources, into a data infrastructure containing curated data in Portuguese that can be integrated into different systems and sources of knowledge in health. In response to this problem, HealFabric was designed and developed, an open and scalable data infrastructure that promotes interoperability between different sources of health information. The core of HealFabric is HealDB, an open-source, structured database in Portuguese. HealDB is generated by a reproducible, layered pipeline that includes data extraction, curation, standardization, and interoperability, following the FAIR principles. In August 2026, HealDB contained over 800.000 records, including medications, active ingredients, drug-disease associations and drug-food interactions, as well as more than 14.000 standardized external identifiers. These identifiers enable interoperability with clinical terminologies, pharmacological classifications, chemical databases, drug databases, biodiversity resources, and repositories of scientific publications. HealFabric and HealDB are publicly available on GitHub and in the Unicamp research data repository, respectively. This thesis presents four main contributions consolidated in HealFabric. The first is to make available a structured and open health database based on official Brazilian sources. The second is to introduce a reproducible pipeline for extracting, curating, and standardizing clinical data from unstructured texts in Portuguese. As a third contribution, a semantic alignment and interoperability approach with international biomedical databases is proposed, based on integration through external identifiers. Finally, difficulties related to health data in Portuguese are discussed, such as the translation of biomedical terminology and the mapping of active ingredients to standardized vocabularies. As an extensible and reusable data infrastructure, HealFabric promotes interoperability between Brazilian Health Information Systems and international biomedical resources through HealDB, contributing to the expansion of the health data ecosystem in Portuguese.
Examination Board
Headlines:
Claudia Maria Bauzer Medeiros IC / UNICAMP
Agma Juci Machado Traina ICMC / USP
Ronaldo dos Santos Mello CTC/UFSC
Islene Calciolari Garcia IC / UNICAMP
Washington Luiz Miranda da Cunha IC / UNICAMP
Substitutes:
Rodolfo de Carvalho Pacagnella FCM / UNICAMP
Sergio Lifschitz CTC/PUC-Rio
Fabio Andre Machado Porto COMAC/LNCC
Institute of Computing
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognizing you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.