Table 3.
Model | Neuropathological covariates | Modules selected | Cell type proportion selected | R 2 | AUC | 95% CI for AUC |
---|---|---|---|---|---|---|
1 | PHF-1 Tau TDP-43 pathology HLA-DR:Iba1 Microvascular lesion Vascular sum score |
NA | NA | 0.117 | 0.666 | 0.540–0.793 |
2 | PHF-1 Tau TDP-43 pathology HLA-DR:Iba1 Microvascular lesion Vascular sum score |
NA | Endothelial cells excitatory neurons oligodendrocytes | 0.299 | 0.789 | 0.680–0.898 |
3 | PHF-1 Tau TDP-43 pathology HLA-DR:Iba1 Microvascular lesion Vascular sum score |
23 | Endothelial cells excitatory neurons oligodendrocytes | 0.675 | 0.923 | 0.866–0.980 |
Models were constructed using combinations of neuropathological covariates, cell type proportions, and co-expression modules to evaluate their impact on psychosis occurrence. Model 1 used only neuropathological covariates. Model 2 provided a modest improvement by incorporating cell type proportion estimations. Model 3, including neuropathological covariates, cell type proportions, and co-expression modules accounted for the largest percent of variance in occurrence of psychosis.