<p>Few studies have empirically examined the hypothesis that age, sex, and season would be acting as effect modifiers in the association between exposure factor and health outcome when the population are simultaneously exposed to multiple air pollutants and meteorological conditions. We couple air pollution and meteorological data with reported mental disorders (MDs) drawn from approximately 1.8 million hospital outpatient visits in Nanjing, China between 2015 and 2019. Subsequent to predicting the illness risks of MDs using deep learning model, SHAP approach quantifies excess risks attributed to the feature being explained. We generally find the effect modifications in the association between air pollutant or meteorological factor and MD risk by comparison of feature-importance estimates for each stratified group. For example, PM₁₀ increases the risk of depression disorder of females by 1.94 (P_2.5-P_97.5: 1.54, 2.34) visit counts, which is significantly higher than that of males [0.25 (−&#xa0;0.01, 0.50)]; the female-male difference in the risk of anxiety disorder posed by precipitation associates with 0.50 (0.31, 0.69) visit counts. The findings reasoned that efforts should address the effect of multiple risk factors that simultaneously interact with each other on MDs, as surrounding air pollution and climate change proceed.</p> Graphical Abstract <p></p>