Learning about risk: Machine learning for risk assessment

Journal article proposing a risk assessment approach based on deep neural networks and illustrating it on an oil & gas drilling rig ‘drive-off’ scenario. The paper discusses how ML can help process and learn from data to support more continuous, data-informed risk management, while also emphasizing limitations and the need for careful model selection and customization. For PSM, it’s a useful reference for integrating ML into hazard/risk analysis: it frames when data-driven models can augment traditional methods (e.g., scenario evaluation, updating likelihoods) and highlights governance considerations needed to avoid misleading confidence in automated risk estimates.