Radiomics-based machine learning model for diagnosing internal abdominal hernias: a retrospective study.

Intraperitoneal hernia is an acute abdominal disease, with complex imaging features and variable clinical manifestations that challenge surgeons and emergency physicians in early disease assessment and streamlined diagnosis and treatment procedures. We retrospectively included patients with internal abdominal hernia between January 2021 and June 2024. Eight machine learning models were constructed, and the classifier with the best performance was selected based on comparative evaluation. The performance of each model was assessed using the area under the curve (AUC), accuracy, and specificity to determine the optimal radiomics-based predictive algorithm. A total of 107 radiomics features were extracted, revealing distinct features between herniated and normal intestines. A predictive model for internal abdominal hernias was constructed based on a machine learning algorithm incorporating 7 different features. The Random Forest model demonstrated superior performance, achieving an AUC of 1, accuracy of 90%, sensitivity of 80%, and specificity of 100% in validation set. Radiomics analysis of internal abdominal hernias provides substantial data support for early disease diagnosis, but it is still a need for validation with a larger sample size in the future.