Health hazards associated with LNG tank truck transportation accidents in China: a text-mining analysis of accident reports and risk-management implications.

LNG tank truck accidents can expose drivers, emergency responders, road users, and nearby populations to cryogenic release, fire, explosion, and oxygen-deficient environments, creating potentially serious health hazards. This study analyses 52 LNG tank truck transportation accident cases reported in China during 2015-2024 to examine whether a reproducible text-mining workflow can complement manual accident review and support health-hazard identification. The reports were segmented into 1,707 analytical text units, of which 1,393 valid units were retained after preprocessing. TF-IDF was used to identify discriminative terms and Latent Dirichlet Allocation was used to derive five latent themes, with topic selection based on perplexity, coherence, and semantic interpretability. Manual semantic validation was then used to distinguish statistical co-occurrence from conceptual or causal relationships. The resulting themes describe accident and response processes, human and management factors, external environmental conditions, equipment and facilities, and institutional and regulatory factors. Accident evidence was further interpreted in relation to documented or technically established pathways leading to frostbite, thermal burns, blast-related trauma, and asphyxiation. The findings demonstrate the value of combining reproducible text screening with expert interpretation for accident learning, health-hazard prevention, and risk management. The study does not estimate population-level health outcomes, health-economic effects, or the health effects of the low-carbon energy transition.