Background . Analysis of the sequence of HR+ HER2– mBC therapy in routine practice is important for the assessment and optimization of approaches to the treatment of breast cancer. Automatic analysis of electronic medical records (EMR) using artificial intelligence technology makes it possible to collect and analyze a large array of patient data. Purpose . Study of drug therapy regimens and schemes in female patients with HR+ HER2– mBC in routine clinical practice in the Russian Federation. Methods . The retrospective data of patients who received mBC therapy in routine clinical practice in eight healthcare facilities of the Russian Federation was extracted using machine learning methods. The data were obtained from the EMRs of patients who received drug therapy from February 1, 2018 to May 31, 2021. Main study objectives: to determine the proportion of patients who received various types of pharmacotherapies (chemotherapy [CT], hormone therapy [HT], targeted therapy [TT]); the number of pharmacotherapy lines, the proportion of patients with resistance and sensitivity to HT. Additional objectives included the determination of demographic characteristics, TNM stages. Ki‑67 proliferative activity index, mutations in the PIK3CA gene, and the presence of a visceral crisis. Results. The analysis included data from 932 patients diagnosed with HR+ HER2– mBC. The average patient age was 58.9 ± 11.3 years. Primary metastatic breast cancer was diagnosed in 172 (18.4%) patients. The bones were the most common localization of the first metastases (32.6%). Visceral crisis was reported in 29 (3.1%) patients. 21% of patients were sensitive to HT, 11.1% were primarily resistant, and 31.9% of patients had secondary resistance. The 353 (37.9%) patients received one therapy line, 269 (28.9%) patients received two lines, 139 (14.9%) patients received three lines, and 81 (8.7%) patients received four therapy lines. The first-line treatment included mono-regimens of HT (antiestrogens – 22.8%, aromatase inhibitors [AI] – 20.9%), chemotherapy mono-regimens (taxanes – 11.7%, pyrimidine analogs – 4.4%). Among the combined regimens, the most common were combined chemotherapy with cyclophosphamide + anthracyclines (14.3%) and HT + TT: aromatase inhibitors + targeted therapy (5.2%). In subsequent lines, the HT monotherapy remained the most frequently prescribed, and the specific gravity of CT increased with each line. The CDK4/6 inhibitors with Fulvestrant (8.6%) and aromatase inhibitors (5.1%) were most often prescribed as part of the combined regimens. Conclusions. The study was the first experience of using AI technology for the retrospective analysis of data of patients with breast cancer in routine practice. The study shows that in routine clinical practice in Russia there are still high frequency of hormone therapy in mono-regime and chemotherapy in the early lines. iCDK4/6 were used in 9.0%, 14.8%, and 10.8% in the first three therapy lines, respectively. The methodology limitations may be associated with the retrospective study design and possible errors when extracting data from the EMR using artificial intelligence.