Spatio-Temporal Modeling and Artificial Intelligence in Identifying Macroterritorial Public Health Risk Clusters

Introduction: Regional disparities in public health have both statistical and spatio-temporal dimensions.

Objective: To identify the spatio-temporal structure of public health risk and to determine whether territorial clusters of disadvantage persist after accounting for contextual and manageable determinant contours.

Materials and Methods: We used a spacio-temporal panel for 2019–2024 including 85 constituent entities of the Russian Federation and indicators grouped by biological, demographic, medical, behavioral, social, and environmental determinants. Spatial weight matrices were constructed using nearest neighbors, distance, and federal district affiliation. We assessed global and local spatial autocorrelation, residual risk clusters, macroregional effects, models with spatially lagged predictors, and smoothed residual risk. Artificial intelligence was used in a limited interpretable form as an algorithmic spatial diagnostic contour for constructing spatial features, detecting clusters, and integrating results with territorial typology. Interpretable artificial intelligence methods were used to automatically construct spatial features, compare transparent regression models, and assess their quality across regional groups. Local clusters were identified using Local Moran’s I.

Results: A distinct spatial structure was identified for life expectancy, all-cause mortality, and homicide mortality. After accounting for contextual and controllable contours, residual spatial autocorrelation persisted for life expectancy, all-cause mortality, and homicide mortality, but virtually disappeared for healthy life expectancy. In terms of life expectancy and homicide mortality, the Siberia–Far East belt of residual risk was identified regarding life expectancy and homicide mortality. Models incorporating spatially lagged predictors improved the explanatory power for life expectancy and homicide mortality; however, the more complex Spatial Durbin Model was not justified as a universal method for all outcomes.

Conclusions: Spatio-temporal diagnostics should be viewed as an independent layer of public health monitoring. It helps distinguish isolated regional disadvantage from macroterritorial risk clusters and supports more evidence-based interregional priority setting.

1. Vasilieva TP, Larionov AV, Russkikh SV, Zudin AB, Vasyunina AE, Rotov VM. Methodological approach to assessing the quality of public health. Zdorov’e Naseleniya i Sreda Obitaniya. 2023;31(11):15-22. (In Russ.) doi: 10.35627/2219-5238/2023-31-11-15-22

2. Dovbiy IP, Degterenko AN. Public health of a region within the coordinates of the sustainable development goals. Vestnik Yuzhno-Ural’skogo Gosudarstvennogo Universiteta. Seriya: Ekonomika i Menedzhment. 2022;16(4):42-53. (In Russ.) doi: 10.14529/em220405

3. Zhukova AK, Silaev AM, Silaeva MV. Spatial analysis of life expectancy in Russian regions. Prostranstvennaya Ekonomika. 2016;(4):112-128. (In Russ.) doi: 10.14530/se.2016.4.112-128

4. Inozemcev ES, Kochetygova OV. Spatial panel analysis of fertility and life expectancy in Russia. Izvestiya Saratovskogo Universiteta. Novaya Seriya. Seriya: Ekonomika. Upravlenie. Pravo. 2018;18(3):314-321. (In Russ.) doi: 10.18500/1994-2540-2018-18-3-314-321

5. Lakman IA, Askarov RA, Timiryanova VM, Askarova ZF. Spatial modeling of mortality of the working-age population of the Republic of Bashkortostan. Zdorov’e Naseleniya i Sreda Obitaniya. 2023;31(12):7-16. (In Russ.) doi: 10.35627/2219-5238/2023-31-12-7-16

6. Tikunov VS, Vatlina TV, Gaidukov VR, Tikunova IN. Methods of spatial analysis of medical and statistical information. Vestnik SGUGIT (Sibirskogo Gosudarstvennogo Universiteta Geosistem i Tekhnologiy). 2023;28(6):67-76. (In Russ.) doi: 10.33764/2411-17592023-28-6-67-76

7. Okunev IYu, Kushnareva AE. Alternative spatial weights matrices: Methodology and application in calculating LISA. Vestnik Sankt-Peterburgskogo Universiteta. Nauki o Zemle. 2023;68(2):390-413. (In Russ.) doi: 10.21638/spbu07.2023.210

8. Anselin L. Local indicators of spatial association LISA. Geographical Analysis . 1995;27(2):93-115. doi: 10.1111/j.1538-4632.1995.tb00338.x

9. Lorant V, Thomas I, Deliège D, Tonglet R. Deprivation and mortality: The implications of spatial autocorrelation for health resources allocation. Soc Sci Med . 2001;53(12):17111719. doi: 10.1016/s0277-9536(00)00456-1

10. Auchincloss AH, Gebreab SY, Mair C, Diez Roux AV. A review of spatial methods in epidemiology, 2000–2010. Annu Rev Public Health. 2012;33:107-22. doi: 10.1146/annurev-publhealth-031811-124655

11. Fayet Y, Praud D, Fervers B, et al. Beyond the map: Evidencing the spatial dimension of health inequalities. Int J Health Geogr. 2020;19(1):46. doi: 10.1186/s12942020-00242-0

12. Elhorst JP. Specification and estimation of spatial panel data models. IRSR. 2003;26(3):244-268. doi: 10.1177/0160017603253791

13. Halleck Vega S, Elhorst JP. The SLX model. J Reg Sci. 2015;55(3):339-363. doi: 10.1111/jors.12188

14. VoPham T, Hart JE, Laden F, Chiang YY. Emerging trends in geospatial artificial intelligence (geoAI): Potential applications for environmental epidemiology. Environ Health. 2018;17(1):40. doi: 10.1186/s12940-018-0386-x

15. Abdel Magid HS, Desjardins MR, Hu Y. Opportunities and shortcomings of AI for spatial epidemiology and health disparities research on aging and the life course. Health Place . 2024;89:103323. doi: 10.1016/j.healthplace.2024.103323

16. Wang Y, Chen X, Xue F. A review of Bayesian spatiotemporal models in spatial epidemiology. ISPRS Int J Geoinf. 2024;13(3):97. doi: 10.3390/ijgi13030097

17. Rabiei R, Bastani P, Ahmadi H, Dehghan S, Almasi S. Developing public health surveillance dashboards: A scoping review on the design principles. BMC Public Health . 2024;24(1):392. doi: 10.1186/s12889-024-17841-2

18. Zaitseva NV, May IV, Kiryanov DA, Goryaev DV, Kleyn SV. Social and hygienic monitoring today: Status and prospects in conjunction with the risk-based supervision. Health Risk Analysis. 2016;(4):4-13. doi: 10.21668/health.risk/2016.4.01.eng

19. Lebed-Sharlevich YaI, Mamonov RA, Yudin SM. On the system of social and hygienic monitoring in the Russian Federation. Rossiyskiy Mediko-Biologicheskiy Vestnik Imeni Akademika I.P. Pavlova. 2025;33(3):475-482. (In Russ.) doi: 10.17816/PAVLOVJ627492

20. May IV, Koshurnikov DN, Galkina OA. Space-time analysis of risk to public health under the exposure to urban noise (on the example of Perm). Gigiena i Sanitariya . 2017;96(1):3539. (In Russ.) doi: 10.47470/0016-9900-2017-96-1-35-39

21. Starshinin AV, Grechushkina NA, Pokusaev AS. Population health index of Russian regions in the context of SDG key indicators. Zdorov’e Megapolisa . 2024;5(3):4-16. (In Russ.) doi: 10.47619/2713-2617.zm.2024.v.5i3;4-16