Biomarker Performance — ROC, AUC, Cutoffs, and Uncertainty
What it does Reads binary reference labels and continuous biomarker scores, then calculates empirical ROC/AUC, a fixedseed stratifiedbootstrap AUC 95% interval, candidate Youden J cutoffs, confusion matrices, sensitivity, specificity, PPV, NPV, accuracy, and Wilson intervals. Quick start Run it in your browser or download the repository and serve it with python3 m http.server 8080. Inputs A CSV with sample, truth, score columns. The truth field accepts positive/negative, case/control, or 1/0. Outputs ROC plot, AUC interval, editablethreshold metrics, sample classifications, and a provenance CSV. Scientific method Uses pairwiseconcordance empirical AUC, Youden J, Wilson score intervals, and a fixedseed classstratified bootstrap. It explicitly notes the optimism introduced when the same data are used to select and evaluate a cutoff. Privacy All calculations run locally. Do not upload confidential data to thirdparty services. Limitations This software is a calculation and exploratory analysis aid. It does not diagnose individuals, select a clinical cutoff, or replace independent validation, professional review, or regulatory decisions.