Table 3. Prediction metrics for a dataset containing 11 DKD samples among a population also containing other glomerular diseases, healthy volunteers, Surine™, and un-biopsied DM-patients.

Number of DKD samples included	Total dataset size (Samples)	Random-chance accuracy	Prediction accuracy	Positive* samples predicted correctly (Sensitivity)	Negative** samples predicted correctly (Specificity)	PPV	NPV
11	150	50% [7.3, 92.7]	93.3%	4/11 = 36.4%	136/139 = 97.8%	57.1%	95.1%

Note:

^*Positive samples here are those with DKD.

^**Negative samples are those without DKD.