Explainability in Graph Neural Networks for Autism Assessment Using fMRI Analysis

• Grant number: 24/09181-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: October 01, 2024
• End date: December 31, 2025
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: André Carlos Ponce de Leon Ferreira de Carvalho
• Grantee: Matheo Angelo Pereira Dantas
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Company: Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC)
• Associated research grant: 20/09835-1 - IARA - Artificial Intelligence in the Remaking of Urban Environments, AP.PCPE

Abstract

Autism diagnosis is made only from clinical analysis, which is often in-accessible. The use of AI could help doctors in improving the diagnosiswith biological data, such as the Functional Magnetic Resonance Imaging(fMRI). However, such complex data usually relies on the use of similarlycomplex black-box models, lacking interpretability and therefore being lessreliable. Therefore, it is important to use the techniques of Explainable AIto uncover the rationales behind the predictions of the models. Within thiscontext, we propose to conduct an investigation on such techniques. Themain goal of the project is to study explainability methods applicable toGraph Neural Networks, which are machine learning architectures designedto operate on graphs, in order to observe neurological patterns for differentsigns of autism. For that purpose, we propose to test the currently availableexplainability algorithms for GNNs on our data and make a comparativeanalysis, using objective performance measures of explainability.