S1 File - Gaussian Mixture Models and Model Selection for [18F] Fluorodeoxyglucose Positron Emission Tomography Classification in Alzheimer’s Disease

Txt A, Use of prediction model. It explains the use of prediction models of TUM data. Mat B, Prediction model of NC against AD using grand mean saved in MATLAB format. Represents the saved prediction model in MATLAB file format. The model is used for NC against AD, and is trained using the grand mean intensity normalization. The brain voxels are divided into 50 bins, which shows the best predictive performance by an internal cross-validation. BIC is used as the model selection method. In short, the model can be denoted as “predictionModel_grandMean_50_NCAD”. The meaning of the following models can be inferred similarly. Mat C, Prediction model of NC against MCI using grand mean saved in MATLAB format. Represents “predictionModel_grandMean_60_NCMCI”. Mat D, Prediction model of MCI against AD using pSMC saved in MATLAB format. Represents “predictionModel_PSMC_110_MCIAD”. Mat E, Prediction model of NC against MCI using pSMC saved in MATLAB format. Represents “predictionModel_PSMC_80_NCMCI”. Mat F, Prediction model of NC against AD using pSMC saved in MATLAB format. Represents “predictionModel_PSMC_50_NCAD”. Mat G, Prediction model of MCI against AD using grand mean saved in MATLAB format. Represents “predictionModel_grandMean_90_MCIAD”. (ZIP)