Work sometimes comes with health risks, such as exposure to asbestos or constant noise. To investigate these risks, work descriptions are collected and standardized by experts using classification systems. However, this process is expensive and demands a lot of time and expertise.
To relieve human experts, researcher Mathijs Langezaal has developed an AI system using 812,552 work descriptions that automatically classifies such descriptions according to various international guidelines. A precision score is assigned to each classification, which experts can use to determine which codes need to be reviewed and potentially improved. This speeds up the process while ensuring accuracy. The AI system outperforms both experts and other available tools.
This means that research on occupational risks can proceed more quickly and accurately, ultimately leading to safer work environments and a reduction in work-related fatalities.