Table 3. Features Identified by Machine Learning Analysis.
| Ranked using SVM weights | Feature | HC (mean) | IDU (mean) | P (univariate) |
|---|---|---|---|---|
| 1 | sCD40L (pg/ml) | 120.2 | 249.5 | 0.0004 |
| 2 | IgG4 (RU) | 192.7 | 308.6 | 0.0310 |
| 3 | CD19+CD20+ (% of lymphocytes) | 3.8 | 7.2 | 0.0036 |
| 4 | TGF-α (pg/ml) | 0.8 | 1.4 | 0.0056 |
| 5 | TNF-α (pg/ml) | 2.6 | 4.1 | 0.0016 |
| 6 | MIP-1β (pg/ml) | 4.4 | 13.2 | 0.0330 |
| 7 | IgD-CD27- (% of 19+20low) | 48.0 | 27.6 | 0.0015 |
| 8 | MonoHex Ceramide d18.1/16:0 (ng/ml) | 0.04 | 0.07 | <0.0001 |
| 9 | gp140_IgM (RU) | 634.9 | 1055.0 | 0.0529 |
| 10 | MonoHex Ceramide d18:1/22:0 (ng/ml) | 0.01 | 0.012 | 0.0269 |