To construct a predictive model for the postoperative recurrence of incisional hernia based on preoperative abdominal CT images, with the aim of assisting hernia surgeons in formulating individualized treatment plans.

A cohort of 528 patients with incisional hernia who were treated at Beijing Chaoyang Hospital between 2016 and 2019, was assembled. Preoperative abdominal CT images underwent standardization, yielding 44,380 images. These images were randomly divided into training and validation sets in a 4∶1 ratio to train and validate a convolutional neural network (CNN) model for predicting incisional hernia recurrence. Model performance was evaluated utilizing indicators including sensitivity, specificity, receiver operating characteristic (ROC) curve, and area under the curve (AUC).

Among the 528 patients who underwent incisional hernia repair, 73 experienced recurrence, resulting in a recurrence rate of 13.8%. The study successfully established a CNN model for predicting postoperative recurrence of incisional hernia, with a validated AUC value of 0.840, sensitivity of 85.2%, and specificity of 68.1%.

The CNN model constructed based on preoperative abdominal CT images has a good ability to predict postoperative recurrence in patients with incisional hernia and has a certain role for hernia surgeons in developing individualized treatment plan.