Development of a Deep Learning Model Integrating Computed Tomography Images and Blood Data for the Diagnosis of Acute Cholecystitis

Background In this study, we aimed to develop and evaluate an artificial intelligence-based diagnostic model for the diagnosis of acute cholecystitis (AC) using non-contrast CT images and clinical data. Methods This retrospective study included 197 patients (100 AC, 97 non-AC) treated between January 2016 and December 2025 at a single center. Patients were randomly divided into training (n = 139) and test (n = 58) datasets. Three models were constructed: an imaging-based deep learning model, a clinical data-based machine learning model, and a hybrid machine learning model integrating deep learning-derived imaging features with clinical data. CT images were preprocessed, and gallbladder regions were segmented. Clinical variables included white blood cell counts and levels of C-reactive protein and liver function markers. Model performance was evaluated using accuracy, precision, recall, specificity, F1 score, and area under the receiver operating characteristic curve (AUC). Statistical comparisons were performed using Welch's t-test and Chi-square test. Results The imaging-based model achieved accuracy 0.741, precision 0.703, recall 0.867, specificity 0.607, F1 score 0.776, and AUC 0.863. The blood-based model achieved accuracy 0.914, precision 0.931, recall 0.900, specificity 0.929, F1 score 0.915, and AUC 0.946. The hybrid model showed the highest performance, with accuracy 0.931, precision 0.882, recall 1.000, specificity 0.857, F1 score 0.938, and AUC 0.996. Conclusions A hybrid model integrating CT imaging and clinical data improved diagnostic performance for AC compared with single-modality models.