Examining The Smooth Effect Of Emergency Medical Services Response Time In Traffic Accidents Based On Semi-parametric Additive Logistic Regression Model

With the acceleration of motorization,the traffic safety problem is becoming increasingly prominent.In traffic safety research,a large number of scholars use statistical models to reveal the complicated influential mechanism of traffic accident related factors on the fatality risk of personnel.T this field,the performance and characteristics of the model would directly affect the accuracy and reliability of the estimated results.Among them,the use of unreasonable statistical may lead to misinterpretations of the results.Therefore,it is of great significance to adopt more reasonable model to mine the fatality effects of different factors in traffic accidents.The paper takes the smooth effect of the emergency medical service response time after a traffic accident as the starting point,and propose a semi-parametric additive logistic model to explore the complex relationship between response time and death risk.Firstly,the system composition of emergency medical service is explained in detail,and then this paper analyzes the influence characteristics of emergency medical service on the risk of death of accident casualties from two aspects,namely prehospital delay and quality of emergency medical service.Secondly,based on the consideration of the complexity of factors that caused death in traffic accidents,a semi-parametric additive logistic model is proposed in this paper by combining the traditional logistic model and the generalized additive model(GAM).The smooth function is used to estimate the effect of continuous variables on fatality,and AIC,BIC,R2 and ROC curve are selected.as the model evaluation criteria.Then the paper selects some data from the Fatality Analysis Reporting System(FARS)as a sample,and conducts descriptive statistical analysis and correlation analysis to determine whether it is suitable for model regression.In the end,the main effects,individual effects and interaction effects of response time are respectively analyzed.In the analysis of main effect,the influence characteristics of individual variables are estimated under the average level of other variables.In the analysis of individual effects,two different forms of linear logistic models and semi-parametric additive logistic regression model are adopted to explore the complex influential pattern of response time on the outcomes,and then the performances of the three models are evaluated.In the analysis of interaction effects,this paper establishes four models with interaction terms between response time and other factors to explore the heterogeneous effects of response time with respect to different statuses of other factors.The results show that the semi-parametric additive logistic model and the other two traditional logistic models consider that night,male,not a motor vehicle occupant and not collision with motor vehicle will increase the risk of death in varying degrees.However,However,the semi-parametric additive logistic model greatly improves the flexibility of the model by introducing a smoothing function,and the model evaluation results show that the overall performance is more superior on deeply mining the smooth influential pattern of response time.The smooth curve of the response time determines two critical moments,namely 5 minute and 17 minute,in which the former represents the fastest decline of the odds of survival and the latter is just the "gold time" for operating rescues.In addition,in the case of longer response time from 17 minutes to 25 minutes,the risk of death showed a downward trend.The relationship between response time and mortality risk is different under different levels of other factors when the response time interacts with other factors.