My Projects

Note

This is a collection of all the public repositories that I have worked on. They are organized by project.

[AI Researcher] Infant Mortality | Entity Match

Predicting Infant Mortality in Brazil using Machine Learning and SUS Data Record Linkage $05/2025$ - $12/2025$

[TCC] Graduation's Final Project | Risk Stratification

As a final goodbye to graduation, I built a risk stratification end-to-end project. The work filled the gap left by the lack of research that used publicly available versions of SIM and SINASC, merged both with Entity Matching techniques, and assessed the risk through clustering.

[etlSUS] Data Pipeline

Working with public healthcare data comes with many problems, such as invalid/impossible row values, lack of organization, missing or renamed cols and so on. Due to all this, I wanted to build a data pipeline from the ground up that would be able to handle all of it and produce high quality data ready for usage.

[KDMiLe Prediction] Rare Event Prediction

Sister project to the previous. This work seeks to evaluate multiple estimators (or as I called it: "the other side of the coin"). Work published at KDMiLe as well.

[KDMiLe Balancing] Data Resampling

With the change of project, came the change of focus. This work has a focus on predicting infant mortality by testing and evaluating multiple resamplers. Work published at KDMiLe.

[AI Researcher] Risk classification | AI for Public Health

Use of Artificial Intelligence for risk classification of Black women in prenatal and childbirth care within the Brazilian context.
$03/2024$ - $02/2025$

[DOHaD Apriori] Association Rules Mining

After receiving the acceptance and very positive review from ICISMED. The team had the idea of diving deeper into association rules, this time conducting a more narrow review of the association between APGAR (health indicator collected at the 5th minute of a newborn's life) and maternal characteristics. The abstract of the research was published at DOHaD.

[FIGO Maternal] Exploratory Analysis | Risk Prediction

Once again had the chance to work with healthcare professionals, this time focusing on risk classification. The dataset used was a result of an improved entity matching pipeline. The abstract of the research was published at FIGO.

[ICISMED Apriori] Association Rules Mining

I had the chance to work with healthcare professionals to extract association rules from the SINASC database. According to them, many of the rules found agreed with the literature. The results were published to ICISMED.

[Josif Linkage] Entity Matching

First work for the project. I used Record Linkage Toolkit to match entities from SINASC (live births) to SIM (mortality). The early results were published at JOSIF.