Formula 1 (F1) race outcome prediction is a complex task due to the dynamic interaction between driver performance, vehicle engineering, weather conditions, circuit characteristics, and strategic decision-making (Patil et al. 2023). Beyond competitive performance, the study emphasizes the wellbeing and safety of drivers, who are exposed to extreme physiological and cognitive stress caused by prolonged high G-forces, elevated cockpit temperatures, dehydration, fatigue, and intense mental workload during races (Brown et al. 2018). The motivation of this research stems from the need for proactive and continuous monitoring approaches capable of supporting driver safety, reducing health-related risks, and improving operational decision-making in high-pressure racing environments, rather than relying solely on post-event assessments (Colangelo et al. 2024). This study aligns with the objectives of Industrial Engineering and Operations Management (IEOM) where the F1 driver is considered a critical component in a high-stakes operational system. By applying human factors, predictive safety analytics, and decision support, the study demonstrates how AI can optimize human performance and minimize risk, mirroring the core IEOM goals of enhancing efficiency and system reliability within the context of human-machine system optimization (Kavitha and Kanishk 2025).

This study presents an AI-based prediction framework for F1 race analytics utilizing 75 years of historical data combined with weather-related data (van Kesteren and Bergkamp 2023). The research focuses on data analysis and Machine Learning (ML)-driven prediction models to support race outcome prediction. The study utilizes historical F1 dataset available on GitHub, covering the years from 1950 to 2025, combined with weather data retrieved from the Open-Meteo Application Programming Interface (API). The datasets were analyzed to identify performance trends, qualifying behavior, pit stop efficiency, and environmental factors influencing race outcomes (Seymour 2026). The proposed approach (depicted in Figure 1) integrates manual preprocessing, feature engineering, Exploratory Data Analysis (EDA), and AI-driven feature selection techniques, including SHapley Additive exPlanations (SHAP) to improve data relevance and model interpretability (FastF1 Documentation 2025). Multiple ML models were evaluated for predicting qualifying positions, pit stop performance, and final race standings. The demonstrated implementation adopts the best model, HistGradientBoosting, to generate race outcome predictions and support data-driven analysis (El Haber et al. 2025).

Figure 1. Logic modelling of the AI-based Formula 1 race results prediction framework.

Keywords

Formula 1, driver’s well-being, machine learning, predictive analytics.