A Bayesian network prediction model for gallbladder polyps with malignant potential based on preoperative ultrasound

Area of Science:

• Gastroenterology
• Medical Imaging
• Predictive Modeling

Background:

• Gallbladder polyps (GPs) require accurate risk stratification to prevent unnecessary surgeries.
• Identifying malignant potential in GPs is crucial for appropriate patient management.
• Preoperative ultrasound plays a key role in assessing GP characteristics.

Purpose of the Study:

• To develop a Bayesian network (BN) prediction model for identifying gallbladder polyps (GPs) with malignant potential.
• To focus the model on GPs with a long diameter measuring 8-15 mm.
• To utilize preoperative ultrasound data for the prediction model.

Main Methods:

• Screening of independent risk factors for malignant potential using chi-squared and logistic regression.
• Development and validation of a BN model using data from 1296 patients across 11 tertiary hospitals.
• Establishing relationships between polyp size and other variables within the BN framework.

Main Results:

• Identified age, number of polyps, and polyp dimensions (long and short diameters) as independent risk factors.
• The BN model demonstrated predictive capabilities with an AUC of 77.38% (training set) and 75.13% (testing set).
• Model accuracy reached 75.58% (training set) and 80.47% (testing set).

Conclusions:

• A Bayesian network (BN) prediction model is accurate and practical for assessing malignant potential in GPs (8-15 mm diameter).
• The model effectively utilizes preoperative ultrasound data.
• This tool can aid clinicians in decision-making regarding cholecystectomy for GPs.