ASSESSING THE IMPACT OF GEOMAGNETIC AND METEOROLOGICAL FACTORS ON THE INCIDENCE OF ACUTE CARDIOVASCULAR DISEASES IN THE IRKUTSK REGION USING MACHINE LEARNING

Significance. The findings of various studies on the impact of geomagnetic and meteorological factors on the incidence of acute cardiovascular diseases remain contradictory. Modern machine learning methods allow for a deeper and more substantiated interpretation of data and assessment of the influence of geomagnetic activity on human health. Purpose. Using machine learning methods, to study the influence of geomagnetic and solar activity and meteorological factors on the incidence of acute cardiovascular diseases in the Irkutsk region in dynamics for the period from 2018 to 2023. Material and methods. The study analyzed data on the incidence of acute cerebrovascular accidents (stroke) and myocardial infarction on a daily basis for the period from 2018 to 2023, along with data on geomagnetic (Kp and Ap indices) and solar (F10.7 and DST indices) activity and meteorological factors, including air temperature and atmospheric pressure. Statistical and analytical methods, as well as the Random Forest method were applied. Results. To examine the pairwise dependence between the number of hospitalizations for myocardial infarction and stroke and geomagnetic and solar activity parameters (Kp, Ap, F10.7, DST), a correlation analysis was performed using the Spearman’s coefficient. The results demonstrated a weak, yet statistically significant positive correlation between the geomagnetic activity indices Ap and Kp and the number of stroke cases in aggregate for three days, as well as a weak negative correlation between the solar activity index F10.7 and the number of stroke cases with a two-day lag after the solar flare. The linear regression models demonstrated the lowest weighted percentage errors when considering the number of acute cardiovascular events within three days following the geomagnetic and solar activity peaks. The correlation analysis revealed weak, yet statistically significant associations between geomagnetic indices and the number of strokes on the third day after geomagnetic disturbances. The RandomForestRegressor method yielded high coefficients of determination, but failed to identify any significant relationship. Models incorporating a three-day aggregate showed better results, suggesting a possible delayed response to geomagnetic fluctuations. Conclusion. The findings indicate that machine learning models provide a more comprehensive analysis and identify trends more accurately than traditional statistical methods. However, no significant association between geomagnetic activity and the frequency of acute cardiovascular diseases was detected. These results underscore the need for a cautious approach to data interpretation and highlight the importance of focusing on established cardiovascular risk factors and the development and implementation of effective public health interventions. Scope of application. The study results can be applied in the development of measures, primarily preventive ones, within the health system. Keywords: Geomagnetic activity; cardiovascular diseases; myocardial infarction (MI); acute cerebrovascular accident (ACVA); machine learning; Random Forest; linear regression; statistical analysis; meteorological factors