 The study investigates the accuracy of ARIMA and LSTM models in predicting changes in in-vehicle, CO2 concentration levels, while also examining the impact of vehicle windows and recirculation ventilation on CO2 accumulation rates. The findings suggest that both models are effective at forecasting changes in a vehicle's interior environment CO2, but ARIMA has better prediction accuracy than LSTM. The research provides theoretical guidance to traffic safety managers for selecting suitable models for predicting in-vehicle, CO2 concentrations and establishing an effective in-vehicle ventilation warning control system. This article was authored by Jehan, Hanlin and Junkai-Chin.