AI-Powered Road Safety: Researchers Develop Accurate Accident Forecasting Model

Researchers from Pacific National University, in collaboration with other institutions, have successfully developed an autoregressive neural network model for predicting road accidents in the Khabarovsk Territory. The study, published in Intellekt. Innovacii. Investicii, aimed to improve road safety by leveraging machine learning techniques to analyze historical accident data and forecast future incidents. The model's accuracy was evaluated against actual accident rates in Khabarovsk Krai from 2015 to 2023, demonstrating its potential to inform decision-making and prevent accidents.

Key Takeaways:

• The researchers developed an autoregressive neural network model to predict road accidents in the Khabarovsk Territory, using machine learning techniques to analyze historical accident data.
• The study aimed to improve road safety by forecasting future incidents and informing decision-making, with a focus on preventing accidents.
• The model was trained and validated using a recurrent neural network, applying scientific methods of statistical modeling of time series and analysis of the feature space of data.
• The study concluded that the prediction model can be adapted to predict the number and types of crashes, participants, and casualties, and its application can improve the quality of road accident prevention and crash avoidance.
• The model's accuracy was evaluated against actual accident rates in Khabarovsk Krai from 2015 to 2023, demonstrating its potential to inform decision-making and prevent accidents.
• Further research will aim to improve the forecasting model by incorporating more information into the input data.

Statistics:

• The study analyzed historical accident data from 01.01.2015 to 30.11.2023 for the Khabarovsk Territory.
• The researchers used a recurrent neural network to model and forecast accident rates, applying an autoregressive approach.
• The model's accuracy was evaluated against actual accident rates, demonstrating its potential to inform decision-making and prevent accidents.
• The study concluded that the prediction model can predict the number and types of crashes, participants, and casualties, and its application can improve the quality of road accident prevention and crash avoidance.

Sources:

• Development of an autoregressive neural network model for predicting accidents in the Khabarovsk Territory. Intellekt. Innovacii. Investicii, 2025, 3():107-120.
• Orenburg State University (publisher of Intellekt. Innovacii. Investicii).
• VerticalNews (news source).
• Journal of Engineering (news source).