The study dealt with the development of a prediction model for the detection of pulmonary hypertension (increased pressure in the pulmonary vessels) in patients with COPD using machine learning. The research team analyzed data from 523 hospitalized patients with COPD and identified 18 non-invasive clinical variables that could predict the development of pulmonary hypertension. Of the eight machine learning algorithms tested, the CatBoost model proved to be the most effective, with an accuracy of 83 percent and a sensitivity of 75.8 percent. The most important indicators for predicting pulmonary hypertension were the diameter of the right ventricle, the diameter of the pulmonary artery, the level of carbon dioxide in the blood, the transverse diameter of the right atrium and the age of the patient. The model uses readily available non-invasive examinations, which could allow earlier detection of the risk of pulmonary hypertension without the need for invasive procedures. The results suggest that this prediction model could improve the early diagnosis and prognosis of patients with COPD complicated by pulmonary hypertension.