Guidelines for setting cut-off scores in AUC (AUC-GUIDE): balancing sensitivity, specificity, and purpose

Area of Science:

• Biostatistics
• Medical Diagnostics
• Health Informatics

Background:

• Determining optimal cut-off scores is crucial for accurate screening and diagnostic test evaluation.
• Area Under the Curve (AUC) from Receiver Operating Characteristic (ROC) curves are common metrics but do not directly identify the best decision thresholds.
• Finding a universally optimal cut-off score is often impractical in real-world scenarios.

Purpose of the Study:

• To present a practical guideline for selecting appropriate cut-off scores in diagnostic and screening tests.
• To integrate Area Under the Curve (AUC) values with sensitivity, specificity, and the intended scientific application.
• To offer a flexible framework for researchers and clinicians to make informed decisions.

Main Methods:

• Developing a guideline based on AUC ranges and performance metrics like sensitivity and specificity.
• Incorporating realistic considerations and scientific priorities into the selection process.
• Utilizing Youden's Index and other techniques to balance performance metrics.

Main Results:

• Recommended strategies for cut-off score selection across various AUC ranges.
• A framework that allows prioritization of sensitivity, specificity, or balanced performance based on study needs.
• Demonstrated practical application of the guideline in test evaluation.

Conclusions:

• The proposed guideline facilitates evidence-based, transparent, and context-sensitive interpretation of diagnostic accuracy results.
• Researchers and clinicians can use this framework to make more robust decisions when selecting cut-off scores.
• Enhances the utility of AUC and other metrics in practical clinical and research settings.