Skip to main content

Table 2 Subgroup analysis. Patients were divided into subgroups based on the following criteria: early (Stages 1–2)/ late (Stages 3-4) CKD stage, younger (under 60)/older (over 60) age, and gender so that each patient was ultimately referenced to one of eight possible different subgroups. The final trained model was implemented on each subgroup

From: Machine learning algorithm for early detection of end-stage renal disease

 

Subgroup size

Positive cases

C- statistics

Sensitivity

Specificity

PPV

NPV

Males ckd S3/S3. Age 60+

1784

164

0.919

0.756

0.931

0.528

0.974

Males,ckd S3/S4, Age 60-

348

44

0.878

0.659

0.908

0.509

0.948

Males ckd S1/S2 Age 60+

1061

16

0.925

0.625

0.983

0.357

0.995

Males ckd S1/S2, Age 60-

559

5

0.968

0.600

0.982

0.231

0.996

Females ckd S3/S4 Age 60+

1862

152

0.918

0.711

0.944

0.529

0.973

Females ckd S3/S4 Age 60-

230

34

0.891

0.765

0.913

0.605

0.957

Females ckd S1/S2 Age 60+

1064

13

0.906

0.765

0.991

0.526

0.997

Females ckd S1/S2 Age 60-

426

10

0.918

0.300

0.993

0.500

0.983