A Comparative Study of Road Traffic Mortality Rates Using Time Series Forecasting Models

Background and Purpose: The increase in road traffic accidents (RTAs) due to urban and suburban expansion and rising vehicle numbers complicates mortality trend analysis. This study aims to compare RTA mortality rates in urban and suburban areas of Golestan Province, north of Iran, using time-series forecasting models. Materials and Methods: This retrospective study examined all RTA data (n=۳۷, ۴۰۹) recorded by the emergency medical service system of Golestan Province from March ۲۰۲۱ to March ۲۰۲۳ for urban and suburban areas. We employed three forecasting models, including multiple logistic regression (MLR), autoregressive integrated moving average (ARIMA), and propensity score matching (PSM), within a time-series data framework, to evaluate their performance in predicting RTA mortality rates in urban and suburban areas and find the demographic factors predicting the mortality rates. We calculated the root mean square error (RMSE) and the mean absolute percentage error (MAPE) to evaluate the prediction accuracy of each model. Results: The survival rate was ۹۸.۷% (n=۳۶۹۳۱) and only ۱.۳% (n=۴۷۸) led to death. Over the two-year period, the RTA mortality rate was significantly higher in suburban areas (۱.۶%) than in urban areas (۰.۸%) (P=۰.۰۰۱). The PSM model outperformed other models with lower RMSE and MAPE for both urban and suburban areas. Age and oxygen saturation (SPO۲) were the significant predictors of RTA mortality rate.  Conclusion: The RTA mortality rate is higher in suburban areas of Golestan Province than in its urban areas. The PSM model provides higher prediction accuracy than the RTA and MLR models for both urban and suburban areas.

کلیدواژه ها:

Time series ، Road traffic accidents (RTAs) ، Forecasting ، Propensity score matching (PSM) ، Autoregressive integrated moving average (ARIMA)