Analysis of trends and forecasting of tuberculosis mortality at the regional level

Abstract

The aim of the study was to analyse trends and forecasting of tuberculosis mortality at the regional level by use of mathematical methods: linear regression and model based on artificial intelligence machine learning and to compare the results of accuracy of prognosis. The analysis of the constructed trends showed that in the Sverdlovsk region over the past 10 years there has been a stable statistically reliable trend towards a decrease in the mortality rate from tuberculosis with an average rate of decrease of -10.5% per year. As of the end of 2022, in the Sverdlovsk region, the studied indicator decreased by 66.9% compared to the baseline level of 2012.The forecast using regression allowed us to obtain values of indicators close enough to the actual ones, however, it is linear and overly optimistic, assuming zero mortality from infection in the coming years, which cannot be reliable due to the sufficient number of tuberculosis patients whose probability of death is not zero. A dynamic simulation model based on machine learning is a more complex and subtle forecasting tool that takes into account many factors, allows you to obtain relevant values of predicted indicators, but it requires increased accuracy which can be achieved by additional training, which will require search for additional factors affecting tuberculosis mortality.

Keywords

• tuberculosis
• mortality
• dynamic simulation model
• machine learning
• artificial intelligence