An innovative diagnostic platform based on endovideocolonoscopy and artificial intelligence for a comprehensive assessment of the functional state of interintestinal anastomoses

Objective. To comprehensively evaluate the potential of integrating artificial intelligence (AI) with endovideocolonoscopy for accurate diagnosis of delayed complications after the formation of interintestinal anastomoses in colorectal surgeries. Materials and methods. The study included 154 patients (81 men, 73 women; mean age 47.2 ± 10.3 years). Since mid-2024, the ChatGPT-4 and DeepSeek-R1 AI algorithms have been used to analyze endoscopic images. Patients were stratified by the types of anastomoses: ileotransverse anastomosis (20.1 %), intussusceptional ileotransverse anastomosis (16.2 %), colostomy anastomosis (15 %), and colorectal anastomosis (48.7 %). The cohort was divided into two groups: control (standard examination) and experimental (using AI). Comparative analysis was performed using the χ² criterion and Student›s t-test; differences were considered statistically significant at p < 0.05. Conclusions. The study demonstrated a clear superiority of the combined diagnostic approach (AI + endovideocolonoscopy) in identifying late complications of interintestinal anastomoses. AI algorithms significantly increased the accuracy of detecting morphological changes, which is confirmed by statistically significant differences (p < 0.05) in sensitivity and specificity compared to standard methods. In particular, hyperplastic changes in the anastomotic zone were observed in 18.8 % of patients with side-to-side ITA (group 1), associated with the absence of the Bauhinia valve. In the group with intussusceptional ITA (group 2), such pathologies were absent, which emphasizes its advantages in preventing mucosal hypertrophy. The use of automated AI analysis (ChatGPT-4, DeepSeek-R1) increased the overall detection of pathologies by 11.7 %, confirming the need to integrate AI into clinical protocols to optimize surgical strategies and reduce the risk of late complications.

Hali tarjima qilinmagan