Research on the influencing factors and mechanism of regional death pattern in China based on functional neural network method

The variation in age-specific death probability is closely linked to demographic, socioeconomic, and geographical factors. The present study employs a functional neural network regression model to examine the influence of these factors on regional death patterns in China, with a specific focus on individuals aged 40 and above, from a nonlinear perspective. In comparison with conventional linear models, this approach is shown to more effectively capture the intricate relationships present in death patterns, thereby enhancing both the predictive performance and the interpretability of the results. Key findings include: (1) Fifteen key factors influencing regional death patterns are identified, with gender and urban-rural status emerging as the most significant. (2) Educational level has a significant impact on death probability in the 40-44 age group. After the age of 45, probabilities are increasingly affected by climate and economic conditions, while healthcare becomes crucial for those aged 60 and above. (3) Some factors exert different levels of influence on death probability across age groups. (4) Interactions between factors, particularly between urban-rural status and other factors, affect model outputs.