Table 3.

Combinations of Western blot parameters for selected pairwise classification problems. Shown cutoffs are designated for the logistic model e^M/(1 + e^M).

Classification: DM versus DN Micro Model: M = –13.44 + 3.22 * APOA1 + 0.87 * B2M + 27.25 * REG1A Cutoff: 0.30, Accuracy: 85 [72, 94], Sensitivity: 83 [63, 95], Specificity: 88 [68, 97]

Classification: DN Micro versus DN Macro Model: M = 27.57 –4.51 * APOA1 –0.47 * B2M –2.14 * CDH1 Cutoff: 0.20, Accuracy: 88 [75, 95], Sensitivity: 100 [86, 100], Specificity: 75 [53, 90]