Diagnostic Testing Terms
| Term | Definition | Example |
| FPR | Proportion of incorrect positive results among those without disease | FPR of SLNB in T1–2 oral squamous cell carcinomas was 29.3% (26) |
| Specificity (1 − FPR) | Probability of negative results among those without disease (true-negative rate) | Specificity of SLNB in T1–2 oral squamous cell carcinomas was 70.7% (26) |
| FNR | Proportion of incorrect negative results among those with disease | FNR of SLNB in T1–2 oral squamous cell carcinomas was 9.8% (26) |
| Sensitivity (1 − FNR) | Probability of positive results among those with disease (true-positive rate) | Sensitivity of SLNB in T1–2 oral squamous cell carcinomas was 90.2% (26) |
| NPV | Probability of not having disease given that test result was negative | NPV in NRG-HN002 for 2-y locoregional control of head and neck was 94.5% (24) |
| PPV | Probability of having disease given that test result was positive | For skull base lesions, PPV was 80%, 60%, and 68.4% and NPV 100%, 83.3%, and 75% for radiologist’s interpretation, SUV cutoff of 2.5, and SUV cutoff of 3.0, respectively (27) |
| ROC curve | Plot of diagnostic tests’ 1 − specificity by sensitivity for different thresholds | Hyun et al. (32) |
| AUC | Measure of how well classifier can differentiate between 2 diagnostic groups | AUC of 0.71 when predicting 1-y overall survival from changes in 18F-FDG uptake after therapy for Ewing sarcoma family of tumors (32) |
18F-FDG PET/CT = 18F-fluorodeoxyglucose PET/CT; SUV = standardized uptake value.