Evaluation of Modified Sequential Probability Ratio Test and Receiver Operating Characteristics Curve in Kidney Disease Diagnosis Sensitivity-Specificity Analysis
DOI:
https://doi.org/10.57233/ijsgs.v11i2.839Keywords:
ASN, OC curve, ROC curve, MSPRTAbstract
This study evaluates the diagnostic performance of the Modified Sequential Probability Ratio Test (MSPRT) and Receiver Operating Characteristic (ROC) curve in kidney disease detection using patient data from Borno, Maiduguri specialized hospital. The analysis compares the efficiency, sensitivity, and specificity of these methods in determining kidney disease progression based on Glomerular Filtration Rate (GFR) and related biomarkers. Results indicate that MSPRT exhibits dynamic decision-making advantages, whereas ROC provides robust threshold-based classification. Results indicate that MSPRT offers adaptive decision-making capabilities, reducing false positives and negatives by continuously updating classification probabilities. However, it requires predefined parameters and can be computationally intensive. In contrast, the ROC curve provides a stable, threshold-based classification method, offering a clear-cut measure of diagnostic performance through the Area Under the Curve (AUC) but lacks real-time adaptability. Findings suggest that while MSPRT can enhance early detection and improve diagnostic efficiency in real-time settings, the ROC curve remains a valuable tool for benchmarking diagnostic accuracy. The ROC curve, on the other hand, provides a graphical representation of diagnostic accuracy by plotting sensitivity against specificity and determining an optimal threshold for disease classification. Both approaches are assessed to identify their strengths and limitations in clinical decision-making. A combined approach integrating the strengths of both methods may provide a more reliable framework for kidney disease classification.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 Author(s)

This work is licensed under a Creative Commons Attribution 4.0 International License.