Researchers from the Universidade de São Paulo (USP) and the Instituto do Câncer do Estado de São Paulo (Icesp) have developed BreastOncoPredict, an Artificial Intelligence (AI) software capable of predicting the risk of recurrence or death in patients with breast cancer.
The study, Ensemble machine learning for predicting breast cancer recurrence and mortality using clinical and hemogram data, was published in npj Breast Cancer, a Nature Portfolio journal.
BreastOncoPredict is among the first AI software tools to incorporate red blood cell count–based parameters to assess the risk of breast cancer recurrence—that is, the return of cancer in the breast or elsewhere in the body (metastasis)—over periods of two and ten years.
The software generates predictions using patients’ clinical information together with data obtained from a complete blood count (CBC), a routine laboratory test that is already performed for other clinical purposes. Because these data are already available, no additional sample collection or testing is required.
Among the expected benefits of BreastOncoPredict is a more accurate prognosis, which may help physicians design more effective treatment strategies. In practice, patients at high risk could receive extended treatment and closer follow-up with additional preventive measures, while low-risk patients could avoid unnecessary interventions.
Another option for predicting the risk of recurrence is the use of breast cancer–specific genomic tests, such as Oncotype DX and MammaPrint. However, these tests require breast tumor tissue samples and cost between US$3,000 and US$4,000 per patient, making them impractical for public healthcare systems in low- and middle-income countries such as Brazil.
According to Brazil’s National Cancer Institute (INCA), approximately 73,610 new cases of breast cancer are diagnosed each year in the country. Based on a conservative estimate, if Brazil’s Unified Health System (SUS) were to provide biopsy-based genomic tests to every woman diagnosed with breast cancer, annual costs would exceed BRL 1 billion.
In this context, adopting BreastOncoPredict within the SUS could provide highly accurate risk prediction without any additional cost.
The researchers expect the innovation to be incorporated into the public healthcare system, beginning with Icesp, where the software was developed. Because the study was retrospective—that is, based on data from Icesp patients who had already been treated—the next step is to conduct a prospective study to validate its clinical use. Subsequently, other public hospitals may also adopt the software.
Beyond the SUS, the software also has international potential. BreastOncoPredict could be implemented in oncology centers across other low- and middle-income countries, where genomic tests are also not widely available and complete blood counts are routinely performed before chemotherapy or surgery.
BreastOncoPredict—currently available in a research-only version—can be accessed through its official website on both computers and mobile devices. The software is available in Portuguese and English.
The study was led by Luciana Rodrigues Carvalho Barros and Patrícia Honorato Moreira, together with Arthur Shuzo Owtake Cardoso, Alexandre Ferreira Ramos, Joaquim Gasparini dos Santos, Rafael de Oliveira, Thomas de Almeida Reichmann, Wesley Antonio Lopes de Lima, Renata Colombo Bonadio, Bruna Salani Mota, Flavia Santoro, and Roger Chammas.
The research was funded by the National Council for Scientific and Technological Development (CNPq), the São Paulo Research Foundation (FAPESP), and the Brazilian Ministry of Health through the Pronon program.
