DxGoals: A Software Tool for Determining and Analyzing Clinically Meaningful Classification Accuracy Goals for...

Area of Science:

• Biostatistics
• Medical Informatics
• Diagnostic Test Evaluation

Background:

• Classification accuracy metrics like sensitivity and specificity are crucial for evaluating diagnostic tests, especially for low prevalence conditions.
• However, establishing clinically meaningful goals for these metrics has been challenging.
• A risk stratification framework was proposed to address this, highlighting the need for a supporting software application.

Purpose of the Study:

• To introduce DxGoals, a freely available R-Shiny software application.
• To enable determination, visualization, and analysis of classification accuracy goals for diagnostic tests.
• To facilitate the assessment of clinically meaningful standalone and comparative diagnostic test performance.

Main Methods:

• DxGoals utilizes Bayes Theorem to translate desired positive predictive value (PPV) and negative predictive value (NPV) into equivalent goals for positive likelihood ratio (PLR) and negative likelihood ratio (NLR).
• The software determines implied goals for sensitivity and specificity based on user-defined prevalence and predictive value targets.
• It also analyzes provided study data to assess goal achievement and facilitates comparative analysis between tests.

Main Results:

• The DxGoals application was illustrated using diagnostic tests for penicillin allergy, ovarian cancer, and cervical cancer.
• Input parameters such as prevalence (p), target PPV (PPV*), and target negative predictive value (NPV*) were informed by clinical management guidelines.
• The software successfully determined, visualized, and analyzed classification accuracy goals for these examples.

Conclusions:

• DxGoals provides a user-friendly platform for setting and evaluating clinically meaningful diagnostic test accuracy goals.
• The application supports both standalone and comparative test evaluations.
• DxGoals is a valuable tool for researchers and clinicians involved in diagnostic test evaluation.