Deriving Clinically Interpretable Cutoffs for Disease Biomarkers

An R package for biomarker evaluation that introduces a two‑stage cutoff framework. Instead of a single threshold, it defines both a rule‑out cutoff (cL) and a rule‑in cutoff (cH), creating three groups: non‑diseased, diseased, and an intermediate gray zone. This approach improves diagnostic interpretability and supports clinical decision‑making.

---

✨ Features

The TwoCutoff package supports reproducible diagnostic pipelines by integrating:

• 📌 adjust_score() -- Gaussian mixture + confounder‑adjusted XGBoost risk modeling\
• 📊 derive_cutoffs_percentile() -- percentile‑based cutoff derivation\
• 📊 derive_cutoffs_sensspec() -- ROC‑based cutoff derivation\
• 📈 evaluate_performance() -- sensitivity, specificity, PPV, NPV metrics\
• 🔍 compare_performance() -- side‑by‑side comparison of cutoff methods\
• 📉 dca_analysis() -- decision curve analysis for clinical utility\
• 🎨 plot_two_cutoff() -- visualization of cutoffs and confusion matrix

---

🛠 Installation

# Install from CRAN (when available) 
install.packages("TwoCutoff")

# Or install the development version from GitHub:
install.packages("remotes")
# remotes::install_github("kumarbhrigu/TwoCutoff")
install.packages("TwoCutoff", repos = c('https://kumarbhrigu.r-universe.dev', 'https://cloud.r-project.org'))