Introduction (Relevance). Colorectal cancer is the third most common cancer and is highly preventable with timely screening. A key organizational challenge remains the structural imbalance between the growing volume of screening programs and the shortage of endoscopists, which cannot be addressed solely through personnel measures. Artificial intelligence technologies (AI) are considered a tool for mitigating the subjective factor and compensating for the physiological degradation of operator attention; however, their organizational and economic integration into the compulsory medical insurance system of the Russian Federation remains insufficiently substantiated. Objective. To compare the diagnostic accuracy of screening colonoscopy results interpretation using four review models—expert assessment, autonomous use of the digital imaging system, a combination of these models, and live review—followed by an economic justification for implementing the digital imaging system in healthcare practice. Materials and Methods. The study was conducted at the Voyno-Yasenetsky Institute of Surgery and the Euromedservice Medical Center. A sample of 200 digital screening colonoscopy records was selected using stratified sampling: 100 examinations with histologically verified target pathology and 100 examinations without pathological changes. The average patient age was 56±6 years. Results. The combined AI + expert model demonstrated the best performance: sensitivity 97.0% (95% CI: 91.5–99.4%), specificity 97.0% (95% CI: 91.5–99.4%), overall accuracy 97.0% (95% CI: 93.6–98.9%), AUC = 0.970. Isolated expert assessment provided sensitivity of 95.0%, specificity 97.0%, accuracy 96.0%, AUC = 0.960. Autonomous use of AI was characterized by a high sensitivity of 82.0% with an unacceptably low specificity of 55.0% and accuracy of 68.5%, AUC = 0.685, due to a significant number of false positive results. Discussion. The performance of the autonomous algorithm excludes its independent use in mass screening settings at the current stage of technology development. The integration of the AI into expert review is statistically significantly superior to both standalone review (p=0.038) and standalone AI (p