Causal discovery can determine the causal rela-tionship between input and output features on available data andperform an accurate analysis of input and output features. Also selectingthe input features of the model based on the strength of the causalrelationship can improve the efficiency of the model calculation andavoid the influence of interfering factors on the prediction results. Dueto the rise of big data, the data collected from various fields arebecoming increasingly complex, and inputting all the data into themodel for calculation will waste a lot of time and have poor interpret-ability. Therefore, causal AI based on causal discovery is graduallyemerging in various fields. For example, Luo et al. [32] analyzed thefactors influencing flight safety through causal discovery and concludedthat unreasonable scheduling of flight support personnel was a signifi-cant cause of flight delays. Zhang et al. [33] utilized Rubin’s potentialoutcome framework to perform the causal inference and developed fourindicators for the influence of disruptions on travel demand, averagetravel speed, and passenger flow distribution to analyze the vulnera-bility of the metro system. Kotoku et al. [34] adopted Direct-LiNGAM tofind causal relationships among health indicators for age groups andgender.