Figure 1
From: A multimodal machine learning model for predicting dementia conversion in Alzheimer’s disease
Performance comparison in AUC of GBM model for modality combinations of each testing set obtained by performing 100 iterations of data shuffling. AUC showed a statistically significant improvement in the modality combination that added image features compared to when only demographic characteristics were used (p-value < 0.05). There was a statistically significant improvement in the AUC when adding MRI image features to the model compared to using the demo + A modality combination (p-value < 0.05). However, there was no statistically significant difference in the AUC between the model using the demo + AN modality combination and the model using the demo + ANV modality combination (p-value = 0.520). GBM;gradient boosting model, demo; demographic characteristic, A;amyloid PET image features, N; T1-weighted image features, V; T2-FLAIR image features.
