Skip to main content
. 2021 Feb 22;13:630382. doi: 10.3389/fnagi.2021.630382

Figure 6.

Figure 6

ROC curve from the support vector machine classifier and the general ROC model for classification of the AD spectrum. The results revealed that the combination of altered rCBF, ALFF, and ReHo in PCC/PCUN had more power in differentiating subjects with ADS from HC (A) MCI and SCD from patients with AD but not (B) SCD from MCI. The classification was able to differentiate disease groups from HC. Specifically, the AUC for the AD patients was 0.978 (95% confidence intervals from 0.942 to 1.000, p < 0.001), the MCI had 0.958 (95% confidence intervals from 0.897 to 1, p < 0.001), and SCD had 0.915 (95% confidence intervals from 0.82 to 1, p < 0.001) (A). Within disease groups, the combination showed a good ability to distinguish AD from MCI (AUC value = 0.933, 95% confidence intervals from 0.855 to 1, p < 0.001) as well as AD from SCD (AUC value = 0.86, 95% confidence intervals from 0.744 to 0.977, p < 0.001), but not MCI from SCD (AUC value = 0.623, 95% confidence intervals from 0.445 to 0.8, p = 0.224) (B). The blue line represents the HC and AD group, the red line represents the HC and MCI group, the purple line represents the HC and AD group, the orange line represents the AD and MCI group, the green line represents the AD and SCD group, the dark blue line represents the MCI and SCD group, and the gray line represents the reference. AD, Alzheimer's Disease; MCI, Mild Cognitive Impairment; SCD, Subjective Cognitive Decline; HC, Healthy Normal; ROC, Receiver Operating Characteristic, AUC, Area Under the Curve; ALFF, Amplitude of Low-Frequency Fluctuation; ReHo, Regional Homogeneity.