The project Codex.AIThe research project, developed within the Applied Mathematics graduate program at the Institute of Mathematics, Statistics and Scientific Computing (IMECC) at Unicamp, won second place in the Goiás Aberto Award for Artificial Intelligence (GO.IA 2026). The research, focused on automating the coding of diagnoses within the Brazilian Unified Health System (SUS), has accumulated R$ 378 in awards and funding for its practical implementation.
With its second-place finish in the award promoted by the Goiás State Research Support Foundation (Fapeg), the solution received R$ 300. The funds will be used to consolidate the platform and begin clinical trials in public hospitals. The system uses language models adapted to clinical Portuguese to reduce the time doctors spend manually filling out the International Classification of Diseases (ICD).
The platform is the result of the doctoral work of researcher Ricardo da Silva Santos, under the guidance of Professor Cristiano Torezzan, from IMECC. The development involves collaboration with the Faculty of Medical Sciences (FCM) of Unicamp and clinical validation carried out at the Center for Comprehensive Women's Health Care (Caism).
Prior to its recognition in Goiás, the work had already received the award for best short paper at the Brazilian Symposium on Computing Applied to Health (SBCAS 2024), with the research "Use of BERT models in Portuguese for predicting ICD codes in a neonatal context". In addition, the project obtained R$ 78 in funding from the Centelha GO Program, a resource that enabled the creation of a startup to transfer the technology from the university to the healthcare market.
The doctoral candidate is scheduled to defend his thesis in August 2026. Currently, two scientific articles derived from the project are under peer review in specialized journals.
