Table 4.
Comparison of methods for predicting conversion from MCI to AD between this study and similar recent studies. First row: MCI-CPS (5-fold), classification after applying SNF method from ADNI-1 dataset. Second row: Raw classifier, model without SNF in ADNI-1. Other rows list performance of compared methods.
| Study | Markers | AUC-ROC | Acc (%) | Sn (%) | Sp (%) |
|---|---|---|---|---|---|
| MCI-CPS (5-fold) | SNP, mRNA expression data, sMRI | 0.83 | 79.20 | 81.25 | 77.92 |
| Raw classifier | SNP, mRNA expression data, sMRI | 0.78 | 76.00 | 77.08 | 75.32 |
| Lu et al. (2018) | PET | - | 81.55 | 73.33 | 83.83 |
| Wei et al. (2016) | sMRI | 0.74 | 66.00 | 55.30 | 75.90 |
| Gao et al. (2020) | sMRI, age | 0.81 | 76.00 | 80.00 | 73.00 |
| Lehallier et al. (2016) | CSF, sMRI, CICS | 0.82 | 80.00 | 88.00 | 70.00 |
| Westman et al. (2012) | sMRI, CSF | 0.76 | 68.50 | 74.10 | 63.00 |
| Zhang et al. (2012) | CSF, PET, sMRI | 0.80 | 73.90 | 68.60 | 73.60 |
| Young et al. (2013) | PET, sMRI | 0.80 | 74.10 | 78.70 | 65.60 |
AUC, area under ROC curve; Acc, accuracy; Sn, sensitivity; Sp, specificity; sMRI, structural magnetic resonance imaging; PET, positron emission tomography; CSF, cerebrospinal fluid; CICS, Clinical Information and Cognitive Scale.