Exploring the risk and predictive study of outdoor air pollutants on the incidence and mortality of HIV/AIDS

Background: The rising incidence of environmental pollution has heightened concerns regarding the impact of pollutant variations on public health.

Methods: Time series analysis models and BP neural network models were utilized to investigate both univariate and multivariate predictions of HIV/AIDS cases. To evaluate the combined effects of pollutants on HIV/AIDS cases, we employed weighted quantile sum (WQS) regression, a quantile-based g-computation approach (Qgcomp) and Bayesian kernel machine regression (BKMR). Additionally, sensitivity analyses were conducted to further validate our findings.

Results: The incidence and mortality rates of HIV/AIDS in Beijing have demonstrated an upward trend, primarily affecting individuals aged 20-35 years, who account for approximately 63.95 % of cases. In the univariate prediction, the parameters that yielded strong predictive performance for the incidence model were as follows: Holt-Winters: α=0.13, β=0.09, γ=0.34. For the mortality model, the parameters indicating good predictive performance were derived from the SARIMA model: (0,1,3) (0,1,2) [12]. The BP neural network model also exhibited robust predictive performance across various configurations of hidden layers (error ∈ [0.096, 1.324]). The WQS model indicated that only NO₂ had a significant effect, with an overall risk effect of the five mixed air pollutants on HIV/AIDS incidence represented as βWQS (95 %CI) = 0.10 (0.02, 0.18). Meanwhile, the Qgcomp model revealed that NO₂ and AQI have hazardous effects on disease incidence, with weights of 0.514 and 0.486, respectively. Additionally, SO₂ was found to have a harmful effect on disease mortality. In the Qgcomp index and BKMR model, the weights of PM₁₀ and PM_2.5 were predominant in the positive weights.

Conclusions: Various time series and neural network models effectively predict the incidence and mortality rates of HIV/AIDS. Additionally, multiple mixed exposure analyses provide further evidence of significant associations between exposure to air pollution mixtures and HIV/AIDS incidence and mortality rates, with PM_2.5 and PM₁₀ being the primary drivers.