Assessing potential of the gain in the life expectancy of population using artificial neural networks

Abstract

Introduction. At present it is especially vital to search for and test new analytical systems that can give a possibility to predict a medical and demographic situational lowing for multifactorial influence exerted by the environment. Our research goal was to establish regional peculiarities and predictive estimates of potential gain in such an important indicator as life expectancy at birth (LEB) depending on changes in socio-hygienic determinants potent of modifying it. To do that, we took data collected in a RF region where the current demographic situation was rather tense against the backdrop of stable economic conditions. Materials and methods. A potential of the gain in LEB was estimated by modelling cause-effects relations between environmental indicators and life-style related ones, or determinants that determined population health. Models were created by using artificial neural networks. Results. Our methodology was proven to be optimal and precise (differences are equal to 0.98%). It can be applied quite successfully to predict a potential gain in LEB at a regional level together with identifying what modifying factors should be considered priority ones. LEB on the analyzed territory (the Perm region) was established to likely grow by 661.6 days by 2024 and reach 73.12 years; by 855.7 days by 2030 and reach 73.65 years if the current trends related to changes in the analyzed determinants persisted and the achievement of target indicators of national projects and regional development programs. In case the relevant targets set within national projects and regional development programs were achieved, this indicator would grow by 661.6 days and reach 73.12 years. The most significant groups of factors that determine LEB on the analyzed territory against the backdrop of stable economic situation include sanitary-epidemiological welfare (working conditions et al.), public healthcare indicators (population provided with sufficient number of doctors), sociodemographic indicators (expenses on social policies), lifestyle factors (the proportion of the population involved in physical culture and sports; consumption of vegetables and fruits; retail sales of alcoholic beverages, etc.). Their contribution to the gain in LEB varies from 51.2 to 228.6 days. Limitations. Limitations of the study include the model being “stationary” due to its training relying on data collected in 2010-2019; use of a specific set of indicators; failure to consider the influence exerted by the current epidemiological processes (the COVID-19 pandemics). Conclusion. We analyzed data collected in an RF region with a rather tense demographic situation and established that by 2024 an adjusted target LEB value would be achieved there if the trend in changes in socio-hygienic determinants recovered to its pre-pandemic levels. Achievement of target LEB values by 2030 requires additional project activities that consider specific regional features and focus on managing priority determinants and reducing mortality among working age population.