Novel Alzheimer's Disease Stating Based on Comorbidities-Informed Graph Neural Networks

Alzheimer's Disease (AD), the most prevalent form of dementia, requires early prediction for timely intervention. Leveraging data from the Alzheimer's Disease Neuroimaging Initiative (ADNI), our study employs Graph Neural Networks (GNNs) for multi-class AD classification. Initial steps involve creating a patient-clinical graph network considering latent relationships among cognitive normal (CN), mild cognitive impairment (MCI), and AD patients, followed by training several GNN-based techniques for building prediction models. Incorporating comorbidity data from electronic health records into the feature set yielded the most effective classification results. Notably, the GNN model with attention mechanisms outperforms state-of-the-art techniques in multi-class AD classification, achieving an accuracy = 0.92 [0.91,0.94], AUC = 0.96 [0.95,0.96], and F1-score = 0.92 [0.91,0.94]. This work highlights comorbidity data's impact on AD classification and suggests its potential to deepen disease understanding.