AI-Assisted Oncology Navigation
Using machine learning and optimization to monitor cancer patient trajectories in real time and flag deviations that call for intervention.
Lead: Guilherme Tegoni Goedert
Cancer care generates enormous volumes of patient data, but most health services only review that data long after a patient's journey has already gone off track — a delayed diagnosis, a missed treatment window, an avoidable emergency visit. By the time these problems surface in a chart review, the chance to intervene early is often already gone.
This project builds software that monitors oncology patient trajectories as they happen, using machine learning to detect when a patient's path through the health system is deviating from expected care patterns. The goal is to catch problems early enough to act on them: prioritizing outreach, closing gaps in follow-up, and helping patients overcome the structural, cultural, and socioeconomic barriers that keep them from receiving timely treatment.
Beyond individual patient navigation, the system is designed to generate the data Brazil's mandatory hospital cancer registries require, but which is rarely connected back to day-to-day patient care. That same data supports service planning and gives partner health units a benchmarking platform for comparing treatments and protocols, and for evaluating the real-world impact of new technologies as they're adopted.
The project builds on OLIM, LABORA's system for structuring and classifying medical records, and is developed in partnership with oncology teams at Grupo Hospitalar Conceição (GHC).
Partners: Grupo Hospitalar Conceição (GHC)
Funding: FINEP and the Brazilian Ministry of Health