A method of evidencebased risk assessment through modeling infant behavior and injury

In Japan, childhood injury prevention is an urgent problem because unintentional injury is the leading cause of death among children, and this trend fails to improve, An injury prevention approach by product modification is especially very important. The risk assessment is one of the most fundamental methods to design safety products. Howcvcr, the conventional risk assessment has bccn carried out subjectively because product makers have poor data on injuries. This paper proposes a new method of evidence-based risk assessment, and describes text mining of child injury data, probabilistic modeling of child behavior, and probabilistic modeling of child injury as basic methods for realizing the proposed method. Child injury data are collected from medical institutions through an injury surveillance system. Text mining of narrative parts of the injury data helps product designers learn types of potential behaviors related to a product and types of injuries due to the product usc. The results of text mining allow product designers to create injury scenarios which depict injurious incidents from the time of product use to injury occurrence. Probabilistic modeling of child behavior based on behavioral observation data allows product designers to compute the probability of the occurrence of the behavior related to the product. Probabilistic modeling of child injury based on the injury data allows product designers to compute the severity of the injury that results from behavior. The risk of the injurious incident scenario is estimated by the combination of the computed probability of the behavior and the computed severity of the injury.