A new Graph Gaussian embedding method for analyzing the effects of cognitive training

Author summary There is still no cure and no effective drug treatment for Alzheimer's disease (AD). Hence,non-pharmacologicalcognitive intervention for patients at early stages of AD has received a lot of attention due to its non-invasive manner, safety, and scalability. Multi-domain interventions targeting memory and non-memory domains simultaneously are urgently needed for an optimal intervention effect on amnestic mild cognitive impairment (aMCI) patients. However, most of the previous multi-domain cognitive intervention studies evaluated the intervention outcomes based solely on neuropsychological assessment or simple characterization of brain anatomical structural changes. This work is the first of its kind to develop a patient-specificquantitative analysisfor the underlyingfunctionalbrain regional activity changes during the multi-domain cognitive intervention process. Specifically, we used fMRI data from 12 patients, who were trained for three months, and we developed the multi-graph unsupervised Gaussian embedding method (MG2G) to analyze these data. We obtained probabilistic changes across all the brain regions, and we found that brain regions in the default mode network and somatosensory/somatomotor hand, fronto-parietal task control, memory retrieval, and visual and dorsal attention systems had relatively large variations during the non-pharmacological training. These fundamental insights could provide effective new biomarkers for monitoring of aMCI cognitive alteration. Identifying heterogeneous cognitive impairment markers at an early stage is vital for Alzheimer's disease diagnosis. However, due to complex and uncertain brain connectivity features in the cognitive domains, it remains challenging to quantify functional brain connectomic changes during non-pharmacological interventions for amnestic mild cognitive impairment (aMCI) patients. We present aquantitativemethod for functional brain network analysis of fMRI data based on the multi-graph unsupervised Gaussian embedding method (MG2G). This neural network-based model can effectively learn low-dimensional Gaussian distributions from the original high-dimensional sparse functional brain networks, quantify uncertainties in link prediction, and discover theintrinsicdimensionality of brain networks. Using the Wasserstein distance to measure probabilistic changes, we discovered that brain regions in the default mode network and somatosensory/somatomotor hand, fronto-parietal task control, memory retrieval, and visual and dorsal attention systems had relatively large variations during non-pharmacological training, which might provide distinct biomarkers for fine-grained monitoring of aMCI cognitive alteration. An important finding of our study is the ability of the new method to capture subtle changes for individual patients before and after short-term intervention. More broadly, the MG2G method can be used in studying multiple brain disorders and injuries, e.g., in Parkinson's disease or traumatic brain injury (TBI), and hence it will be useful to the wider neuroscience community.