This paper presents an Artificial Intelligence–driven system to enhance occupational safety management within an industrial context. Designed and deployed at Bosch Braga, the solution integrates seamlessly into the Adaptive Business Intelligence (ABI) platform and addresses the automatic processing, analysis, and prediction of safety-related events. The system comprises three core AI modules: (i) a text-mining model for the automated Classification of incident descriptions; (ii) an anomaly-detection module that jointly analyses tabular and textual features to identify inconsistent or atypical records; and (iii) a risk-scoring model that estimates the daily probability of accident occurrence based on historical data. The system was validated using real operational datasets, demonstrating substantial gains in efficiency and data quality. Results show a reduction of more than 50% in manual analysis effort, along with consistent categorization and improved reliability of safety records. Although predictive performance remains constrained by the scarcity of near-miss reporting, the findings confirm both the practical utility and the feasibility of integrating AI into occupational risk-management processes. Overall, the ABI platform supports a shift from reactive to predictive safety practices and establishes a scalable foundation for continuous improvement in industrial Health, Safety, and Environment (HSE) operations.
Pires et al. (Thu,) studied this question.