[The study of impact of medical social factors on the level of life expectancy in Russia: the regressive model]

The science

Original title

[The study of impact of medical social factors on the level of life expectancy in Russia: the regressive model].

Authors

Khabriev RU, Skopintseva ME, Kotlovskiy MY

Journal

Problemy sotsial'noi gigieny, zdravookhraneniia i istorii meditsiny

Published

Oct 29, 2026

Study type

Study · in humans

PMID

42954197 ↗

DOI

10.32687/0869-866X-2026-34-5-1173-1180 ↗

Abstract

The life expectancy is not only an indicator of achievement of the National development goal of the Russian Federation, but also a complex indicator affected by the number of factors.The purpose of the study is quantitative evaluation of relationship and impact of medical social factors on population life expectancy of the Subjects of the Russian Federation.The study was carried out in several stages. At the actual stage the resultant linear regression model was obtained. The key determinants of longevity were verified and their relative importance was assessed using three complementary methods of evaluation of importance of characteristics (analysis of modules of standardized regression coefficients, calculation of global importance on the basis of the Shapley Additive Explanations, permutation importance analysis carried out for additional verification of the ranking of social medical determinants of life expectancy). The comprehensive assessment of descriptive and statistical reliability of the model was carried out, as well as its predictive applicability was determined. The model demonstrated high explanatory capacity since determination coefficient made up 0.811 testifying that more than 81% of the inter-regional variation of life expectancy was explained by included predictors. The additional quality metrics (mean absolute error made up 0.858 years, median absolute error - 0.658 years and mean absolute percentage error - 1.2%) confirmed high accuracy of the model. The ranking by module of β-coefficients established clear hierarchy of determinants. The mortality rates played leading role, especially from external causes (18.1% of total contribution) and circulatory diseases (14.6%). The evaluation of resumptive capacity of the model using 5-fold cross-validation confirmed its potential not only for descriptive analysis of relationships but also for prognostication. The obtained regression model has the potential to be applied in monitoring, planning and evaluation of efficiency of regional health care policy.

Read it on PubMed ↗