Skip to main content

Table 3 Utilization of the PDxBr training and validation models with cutoff to stratify patients into high- and low-risk recurrence groups

From: Development and validation of an AI-enabled digital breast cancer assay to predict early-stage breast cancer recurrence within 6 years

 

Events

Censored

Total

Training model

 *Risk score ≥ 58

140

295

435

 *Risk score < 58

80

1044

1124

Training performance metric

 

Performance (95% CI)

 

 Sensitivity

 

0.715

 

 Specificity

 

0.772

 

 PPV

 

0.319

 

 NPV

 

0.948

 

 Confidence interval

 

0.787 (0.764, 0.808)

 

Validation model

 Risk score ≥ 58

39

104

143

 Risk score < 58

30

343

373

Performance metric

 

Performance (95% CI)

 

 Sensitivity

 

0.605

 

 Specificity

 

0.768

 

 PPV

 

0.242

 

 NPV

 

0.941

 

 Confidence interval

 

0.755 (0.717, 0.792)

 
  1. PPV positive predictive value, NPV negative predictive value
  2. *Model cutoff of 57.77 rounded to 58