Skip to main content
. 2020 Jul 3;17(12):1789–1797. doi: 10.1080/15476286.2020.1783487

Figure 5.

Figure 5.

The multivariate regression model with circRNA candidates in predicting (A) CDR and (B) CERAD state. The reduced model included covariates sex, age of death, batch and the first three principal components calculated from the circRNA expression level. The full model further included the five significantly DE circRNAs with the highest average AUC in the univariate regression models (the five circRNAs used in predicting CDR state were shown in Supplementary Fig 6, the five used in predicting CERAD were not shown)