Deep learning predicts cognitive ability from brain activity and controls
MAC-Net: A Multi-Task Deep Learning Framework for Modeling Cognitive Function From Task-Based fMRI
Artificial Intelligence
Summary
Measuring how well the brain is working by looking at brain scans is hard because other factors like age or the scanner used can confuse the results. The authors created a new deep learning method called MAC-Net that separates brain activity signals from these other factors to better predict thinking skills. They tested their approach on thousands of teenagers’ brain scans and showed it works better than older methods. Their method also identified key brain regions involved in memory and decision-making tasks.
What this means in practice
- •For clinical neuroimaging teams: Improve individual cognitive assessment from fMRI scans by separating brain signals from demographic and scanner noise using MAC-Net.
- •For brain-computer interface developers: Use a covariate-aware neural network to extract cognitive state features from task-related brain activity for adaptive brain-device interactions.
Authors
Md. Tanvir Rahman, Nabil Anan Orka, Asaduzzaman Khan, Mohammad Ali Moni
Abstract
Objective cognitive assessment from neural signals supports neurorehabilitation, but individual-level prediction from task-based fMRI (tfMRI) remains difficult because neural features coexist with substantial demographic and scanner-related variation. We present the Multi-task Activation and Contrast Network (MAC-Net), a covariate-aware deep learning framework for modeling individual cognitive function from regional tfMRI. By isolating tfMRI features into a dedicated neural pathway and restricting participant variables to a terminal late-fusion pathway, MAC-Net prevents dominant covariates from suppressing high-dimensional clinical representations during feature learning. Evaluating baseline data from 6,500 Adolescent Brain Cognitive Development Study participants under family-aware cross-validation, MAC-Net was benchmarked against linear models, random forests, and alternative deep architectures. The N-back plus Monetary Incentive Delay configuration achieved $R^{2}$ values of 0.174, 0.238, and 0.277 for fluid, crystallized, and total cognition, outperforming covariate-only baselines (0.178) and alternative deep models (0.217). N-back was the most informative paradigm, whereas incorporating the Stop Signal Task marginally degraded performance. Feature attributions via Integrated Gradients, DeepLIFT, and Input Gradient were highly concordant, localizing working-memory-related frontal, parietal, and cingulate regions. These findings demonstrate that covariate-aware multi-task modeling yields reproducible cognitive-function estimations, establishing a robust neural engineering framework for clinical translation.