Abstract
BACKGROUND: Accurately predicting the progression of mild cognitive impairment (MCI) to Alzheimer's disease (AD) is a challenging task, which is crucial for helping develop personalized treatment plans to improve prognosis. PURPOSE: To develop new technology for the early prediction of AD using artificial intelligence and cortical features on MRI. METHODS: A total of 162 MCI patients were included from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. By using a 3D-MPRAGE sequence, T1W images for each patient were acquired. All patients were randomly divided into a training set (n = 112) and a validation set (n = 50) at a ratio of 7:3. Morphological features of the cerebral cortex were extracted with FreeSurfer software. Network features were extracted from gray matter with the GRETNA toolbox. The network, morphology, network-clinical, morphology-clinical, morphology-network and morphology-network-clinical models were developed by multivariate Cox proportional hazard model. The performance of each model was assessed by the concordance index (C-index). RESULTS: In the training group, the C-indexes of the network, morphology, network-clinical, morphology-clinical, morphology-network and morphology-network-clinical models were 0.834, 0.926, 0.915, 0.949, 0.928, and 0.951, respectively. The C-indexes of those models in the validation group were 0.765, 0.784, 0.849, 0.877, 0.884, and 0.880, respectively. The morphology-network-clinical model performed the best. A multi-predictor nomogram with high accuracy for individual AD prediction (C-index = 0.951) was established. CONCLUSION: The early occurrence of AD could be accurately predicted using our morphology-network-clinical model and the multi-predictor nomogram. This could help doctors make early and personalized treatment decisions in clinical practice, which showed important clinical significance.