An App model that utilizes a logistic regression algorithm for predicting choledocholithiasis: A prospective clinical...

Area of Science:

• Gastroenterology
• Medical Diagnostics
• Predictive Modeling

Background:

• Current criteria for diagnosing choledocholithiasis (CL) have inaccurate diagnostic yield.
• The American Society for Gastrointestinal Endoscopy (ASGE) provides risk stratification criteria for CL.
• Accurate CL diagnosis is crucial for appropriate patient management.

Purpose of the Study:

• To develop and evaluate a logistic regression model for predicting CL diagnosis.
• To improve diagnostic accuracy in patients categorized as intermediate or high risk for CL.
• To compare the model's performance against existing ASGE criteria.

Main Methods:

• An analytic, observational, cross-sectional study was conducted.
• A logistic regression model was developed and evaluated in 148 patients with intermediate or high risk for CL.
• Receiver operating characteristic (ROC) curve analysis and Endoscopic retrograde cholangiopancreatography (ERCP) as the gold standard were utilized.

Main Results:

• The logistic regression model achieved an Area Under the Curve (AUC) of 0.68 for the overall cohort.
• In intermediate-risk patients, the model showed an AUC of 0.72 and a Positive Predictive Value (PPV) of 70%.
• In high-risk patients, the model demonstrated an AUC of 0.78 and a PPV of 89%.

Conclusions:

• The developed logistic regression model shows improved prediction of CL compared to ASGE criteria in intermediate and high-risk patients.
• The model can effectively guide clinical decision-making for patients with suspected CL.
• Further validation may enhance its clinical utility in diagnosing choledocholithiasis.