Table 3.
A minimal set of markers that are predictive of MCI-to-AD progression. A generalized linear model (GLM) classifier with L1 regularization was utilized to identify a small set of minimally correlated features that can predict MCI-to-AD progression.
| CSF | PIB-PET | FDG-PET | SMRI | Genetic | Demographics |
|---|---|---|---|---|---|
| Total-Tau | COMPOSITE REFNORM | Angular-Left | Cortical Thickness Average of Insula | CR1 | AGE |
| Amyloid Beta | TEMPORAL | Temporal-Left | Cortical Thickness Average of SuperiorFrontal | ||
| Phosphorylated-Tau | Cingulum-Bilateral | Cortical Thickness Average of InferiorParietal | |||
| Temporal-Right | Cortical Thickness Average of Parahippocampal | ||||
| Cortical Thickness Average of MedialOrbitofrontal | |||||
| Cortical Thickness Average of CaudalAnteriorCingulate | |||||
| Cortical Thickness Average of SuperiorParietal | |||||
| Cortical Thickness Average of IsthmusCingulate | |||||
| Cortical Thickness Average of ParsTriangularis | |||||
| Cortical Thickness Average of PosteriorCingulate | |||||
| Hippocampal Subfield Volume of Subiculum |
Those features are categorized here based on their feature modality.