Conventional analyses of resting-state fMRI data often fail to capture nonlinear features that are important for the early detection of Alzheimer’s disease. Here, the authors develop a surrogate model using reservoir computing and compressed sensing to enhance fMRI data interpretation, achieving classification accuracies up to 87% and identifying altered dynamics in cortical regions involved in executive control, working memory, emotional regulation, spatial cognition, and self-referential processing.
- Qing Li
- Zijian Wang
- Jinghua Xiao