On MCC for Medical Diagnostics: Measure of Accuracy and Optimal Cut-off Point Selection Under Tree Ordering of Disease Classes.
Faculty Mentor
Dr. Hani Samawi
Location
Russell Union Ballroom
Type of Research
Completed
Session Format
Poster Presentation
College
Jiann-Ping Hsu College of Public Health
Department
Biostatistics, Epidemiology, and Environmental Health Sciences
Abstract
Accurate differentiation between diseased and non-diseased states is a cornerstone of clinical diagnostics, especially in complex conditions involving multiple disease stages or subtypes. Establishing optimal cut-off points, or thresholds that determine test result classifications, is essential for improving diagnostic precision and clinical decision-making. This paper introduces the Tree Ordering of Disease Subtypes-based Matthews Correlation Coefficient (TMCC), an innovative extension of the traditional Matthews Correlation Coefficient (MCC), designed specifically to evaluate diagnostic performance and identify optimal cut-off points in the context of multi-subtype diseases exhibiting tree or umbrella ordering. Unlike traditional accuracy metrics that may emphasize only sensitivity or specificity, TMCC incorporates all components of the confusion matrix, true positives, false positives, true negatives, and false negatives, offering a holistic and unbiased assessment of test performance. A key advantage of TMCC lies in its robustness to class imbalance, ensuring that negative outcomes and the reliability of negative test results are not undervalued. This makes TMCC particularly valuable in diagnostic settings with varying disease prevalence or when rare subtypes are present. Through extensive simulation studies, TMCC demonstrated comparable or superior performance relative to established metrics across a range of diagnostic scenarios. Additionally, the application of TMCC to real-world clinical data further validated its utility, showcasing its capacity to deliver balanced and context-sensitive evaluations of diagnostic accuracy. TMCC's adaptability and comprehensive framework position it as a powerful tool for modern diagnostic testing, particularly in evolving public health landscapes where disease patterns and prevalence rates are dynamic.
Program Description
.
Start Date
4-23-2026 2:00 PM
End Date
4-23-2026 4:00 PM
Recommended Citation
Gakpo, Jacob O., "On MCC for Medical Diagnostics: Measure of Accuracy and Optimal Cut-off Point Selection Under Tree Ordering of Disease Classes." (2026). GS4 Student Scholars Symposium. 150.
https://digitalcommons.georgiasouthern.edu/research_symposium/2026/2026/150
On MCC for Medical Diagnostics: Measure of Accuracy and Optimal Cut-off Point Selection Under Tree Ordering of Disease Classes.
Russell Union Ballroom
Accurate differentiation between diseased and non-diseased states is a cornerstone of clinical diagnostics, especially in complex conditions involving multiple disease stages or subtypes. Establishing optimal cut-off points, or thresholds that determine test result classifications, is essential for improving diagnostic precision and clinical decision-making. This paper introduces the Tree Ordering of Disease Subtypes-based Matthews Correlation Coefficient (TMCC), an innovative extension of the traditional Matthews Correlation Coefficient (MCC), designed specifically to evaluate diagnostic performance and identify optimal cut-off points in the context of multi-subtype diseases exhibiting tree or umbrella ordering. Unlike traditional accuracy metrics that may emphasize only sensitivity or specificity, TMCC incorporates all components of the confusion matrix, true positives, false positives, true negatives, and false negatives, offering a holistic and unbiased assessment of test performance. A key advantage of TMCC lies in its robustness to class imbalance, ensuring that negative outcomes and the reliability of negative test results are not undervalued. This makes TMCC particularly valuable in diagnostic settings with varying disease prevalence or when rare subtypes are present. Through extensive simulation studies, TMCC demonstrated comparable or superior performance relative to established metrics across a range of diagnostic scenarios. Additionally, the application of TMCC to real-world clinical data further validated its utility, showcasing its capacity to deliver balanced and context-sensitive evaluations of diagnostic accuracy. TMCC's adaptability and comprehensive framework position it as a powerful tool for modern diagnostic testing, particularly in evolving public health landscapes where disease patterns and prevalence rates are dynamic.
