Table 3.
FCBF-based multivariable logistic regression analysis for investigating the effect of gender on lymphoma development.
| Prominent Feature * | Regression Coefficient | Odds Ratio | p-Value | CI Low | CI Upper |
|---|---|---|---|---|---|
| Lymphadenopathy | 1.869 | 6.549 | 0.0003 ** | 2.456 | 17.482 |
| SGE | 0.689 | 2.006 | 0.129 | 0.853 | 4.724 |
| Anti-La | 0.682 | 1.989 | 0.11 | 0.884 | 4.477 |
| Female Gender | −1.119 | 0.332 | 0.011 ** | 0.148 | 0.742 |
| Low C4 (<20 mg/dl) | 0.465 | 1.599 | 0.337 | 0.629 | 4.069 |
| Monoclonal gammopathy | 0.537 | 1.728 | 0.512 | 0.353 | 8.592 |
* The strongest potentially independent variables identified by the fast-correlation-based feature selection (FCBF) algorithm to build the logistic regression model, after analyzing the following initial features included in the dataset: ethnicity, gender, disease duration (onset), dry mouth, dry eyes, anti-Ro, anti-La, Rheumatoid Factor, focus score, germinal centers in biopsy, monoclonal gammopathy, SGE (Salivary Gland Enlargement), lymphadenopathy, low C4, dry skin, chronic fatigue, arthralgias-myalgias, arthritisis, Raynaud’s phenomenon, palpable purpura, vasculitic ulcer, myositis, peripheral neuropathy, CNS involvement, liver-autoimmune hepatitis, liver-PBC (Primary Biliary Cirrhosis), interstitial lung disease, interstitial renal disease, kidney/glomerulonephritis, heart valvular disease, cryoglobulinemia, and lymphoma. ** < 0.05 (95% confidence interval): independent lymphoma-associated features revealed by the logistic regression model.