Background. Prediction of the new coronavirus infection (COVID-19) spread is important to take timely measures and initiate systemic preventive and anti-epidemic actions both at the regional and state levels to reduce morbidity and mortality.Objective: to develop a model for short-term forecasting of COVID-19 cases and deaths in the Russian Federation.Material and methods. The data for the model training were collected from the Stopcoronavirus.rf and Johns Hopkins University portals. It included 13 features to assess the infection dynamics and mortality, as well as the rate of morbidity and mortality in different countries and certain regions of the Russian Federation. The model was trained by the CatBoost gradient boosting method and retrained daily with updated data.Results. The forecast model of COVID-19 cases and deaths for the period of up to 14 days was created. The Mean Absolute Percentage Error (MAPE) estimate of the model’s accuracy ranged from 2.3% to 24% for 85 regions of the Russian Federation. The advantage of the CatBoost machine learning method over linear regression was shown using the example of the Root Mean Square Error (RMSE) value. The model showed less error for regions with a large population than for less populated ones.Conclusion. The model can be used not only to predict of the pandemic of the novel coronavirus infection but also to control and assess the spread of diseases from the group of new infections at their emergence, peak incidence, and stabilization period.

The course of the COVID‑19 pandemic imposes a significant burden on healthcare systems, including on primary care, when it is necessary to correctly suspect and determine further management. The symptoms non-specificity and the manifestations versatility of the COVID‑19 impose difficulties in identifying suspicions. To improve the definition of COVID‑19 symptom checkers and medical decision support systems (MDSS) can potentially be useful. They can give recommendations for determining the disease management. The scientific analysis shows the manifestations versatility and the occurrence frequency COVID‑19. We structured the manifestations by occurrence frequency, classified them as “large” and “small”. The rules for their interaction were determined to calculate the level of suspicion for COVID‑19. Recommendations on patient management tactics were developed for each level of suspicion. NLP models were trained to identify the symptoms of COVID‑19 in the unstructured texts of electronic health records. The accuracy of the models on the F-measure metric ranged from 84.6% to 96.0%. Thus, a COVID‑19 prediction method was developed, which can be used in symptom checkers and MDSS to help doctors determine COVID‑19 and support tactical actions.