Generative forecasting of brain activity enhances Alzheimer's classification and interpretation
This work addresses Alzheimer's Disease classification for neuroscience and healthcare, but it appears incremental as it applies existing methods (LSTM and Transformer) to a new domain-specific dataset.
The study tackled the challenge of classifying Alzheimer's Disease using limited rs-fMRI data by employing generative forecasting as data augmentation, resulting in enhanced classification performance with concrete gains, though specific numbers were not provided in the abstract.
Understanding the relationship between cognition and intrinsic brain activity through purely data-driven approaches remains a significant challenge in neuroscience. Resting-state functional magnetic resonance imaging (rs-fMRI) offers a non-invasive method to monitor regional neural activity, providing a rich and complex spatiotemporal data structure. Deep learning has shown promise in capturing these intricate representations. However, the limited availability of large datasets, especially for disease-specific groups such as Alzheimer's Disease (AD), constrains the generalizability of deep learning models. In this study, we focus on multivariate time series forecasting of independent component networks derived from rs-fMRI as a form of data augmentation, using both a conventional LSTM-based model and the novel Transformer-based BrainLM model. We assess their utility in AD classification, demonstrating how generative forecasting enhances classification performance. Post-hoc interpretation of BrainLM reveals class-specific brain network sensitivities associated with AD.