Development and validation of an artificial intelligence platform with continuous learning for precision Oncology

• Grant number: 22/12991-0
• Support Opportunities: Regular Research Grants
• Start date: February 01, 2023
• End date: January 31, 2025
• Field of knowledge: Biological Sciences - Genetics - Human and Medical Genetics
• Principal Investigator: Israel Tojal da Silva
• Grantee: Israel Tojal da Silva
• Host Institution: A C Camargo Cancer Center. Fundação Antonio Prudente (FAP). São Paulo , SP, Brazil
• City of the host institution: São Paulo
• Associated researchers: Alexandre Defelicibus ; Diana Noronha Nunes ; Dirce Maria Carraro ; Emmanuel Dias-Neto ; Evandro Sobroza de Mello ; Felipe José Fernandez Coimbra ; Mariangela Ottoboni Brunaldi ; Rafael Andres Rosales Mitrowsky ; Rodrigo Drummond Couto Duarte ; Zhao Liang
• Associated scholarship(s): 23/06867-8 - Development and validation of an artificial intelligence platform with continuous learning for precision oncology., BP.TT

Abstract

The use of molecular biomarkers has great potential to drive precision oncology. Although promising, they are not yet widely available in the clinical routine of most medical centers and their high cost can make them unaffordable for most patients. Thus, the development of new fast and low-cost approaches becomes increasingly necessary. In this context, the use of modern techniques of Artificial Intelligence (AI) applied to the analysis of complex data represents an opportunity to expand the evaluation of biomarkers to a large number of patients. This project intends to integrate results of molecular tests, obtained from assays considered gold standard, together with digitized images with the morphology of the tumor tissue stained with Hematoxylin-Eosin (H&E). In the end, it is expected to build an intelligent system of continuous learning capable of providing a panel of digital analogues to biomarkers, whether diagnostic, predictive or prognostic. Importantly, we will use explainability principles in order to gain understanding of which (groups of) known or new parameters have a high influence on the predictions obtained from AI models. Finally, we will build a valuable open access database, embed a panel of models in a computational interface and make it available to the medical and scientific community in order to support the generation of hypotheses and the validation of findings that can contribute to the areas of health and AI, with an emphasis on Oncology. (AU)