The need for noninvasive investigative techniques for solitary lung masses/nodules is increasing with more and more gained experience in surgical treatment of benign lung tumors, including hamartomas.Current radial diagnostic capabilities can be com-ple mented by the introduction of a radiomics-based artificial intelligence (AI) system to recognize pulmonary hamartoma in routine practice, which is currently a pro-mising direction in radiology. Objective. To sum marize the data on the results and conditions of applica tion of radiomics-based AI in the identification of hamartomas from the structure of pulmonary nodules. Materials and methods. The review was performed according to the PRISMA standard (Preferred Reporting Items for Systematic Reviews and Meta-Analyses). Full-text articles published between 2020 and 2024 were included. Articles were searched independently in CyberLeninka (КиберЛенинка), eLibrary, Google Scholar, and PubMed/MEDLINE databases for key terms without language restrictions: “pulmonary hamartoma”, “radiomics”, “machine learning”, “artificial intelligence”. Results. The final analysis included 4 scientific publications on the qualitative indicator. The article presents: description of training and validation datasets; machine learning algorithms; parameters of diagnostic efficiency of the radiomic model (ROC AUC — area under the ROC curve, accuracy, sensitivity and specificity). Radiomic features that distinguish hamartoma from lung malignancies (carcinoid, adenocarcinoma) were the subject of all publications studied. AI application analysis demonstrated that the derived radiomic models in the studies achieved very good (0.836) and excellent (0.942; 0.96) quality scores on the internal test dataset according to the ROC AUC metric, as well as good (0.76) quality scores on the validation test. Conclusion. Noninvasive differential diagnostics of nodules and masses in the lungs is extremely important nowadays. Allocation of obviously benign masses (hamar tomas) from the structure of all found nodules according to CT data will help reduce the number of unnecessary verifying surgeries, and will bring direct economic effect. Scientific publications on this problem and their results confirm the prospectivity of studying the possibilities of AI and radiomic analysis of CT-images in detection of benign pulmonary neoplasms (hamartomas). Validation of the developed models is recommended for all studies, but the latter may be limited due to the small number of cases of verified pulmonary hamartomas in a single institution, which encourages researchers to establish a multicenter base for more effective training of AI based on radiomics and verification of sensitivity, specificity and diagnostic accuracy.