Database Construction and Data Mining for Uncovering the Mechanism of Human Injury in Seismic Damage Buildings

(1)研究背景揭示破坏性地震破坏建筑物中人体损伤的机理是预防死亡发生的根本解决方法。然而，传统的研究直接从建筑物倒塌的数量来估计死亡人数。这种荒谬的情况是由于建筑结构工程师和灾害医学研究人员在研究领域没有联系，他们一直在各自独立的组织下推进研究。(2)项目目标：跨学科探讨地震工程与灾害医学的最新研究成果。其次，通过对数据库进行回归分析和一些统计工作，试图提出死亡人数估算模型，从而为减少人员伤亡提供具体措施。本项目以2004年伊朗Barn Earth…More地震为砖石结构倒塌地震，1999年土耳其koocaeri地震为钢筋混凝土结构倒塌地震，1995年日本神户地震为木结构倒塌地震。(3)项目实施启动会议：为在第一阶段与工程领域和医学领域的研究人员讨论3年的研究计划，日美联合开展了跨学科的研究交流研讨会。德克萨斯大学健康科学中心的合作。研讨会以公开的方式举行，并通过互联网同时向全世界发布研讨会的情况。数据库建设：在研讨会讨论的基础上，从震害建筑死亡发生的角度重新定义了建筑损伤的定义，目的是建立数据库，明确死亡发生机制。新的定义被可视化为建筑物损坏的模式图。在此基础上，利用高织里地震和神户地震的死亡人数和建筑物损坏信息重建了该数据库。此外，我们还通过增加各研究机构收集的有关数据来整合数据库。建模估算死亡人数及其应用：利用事件树分析法对神户地震数据库的死亡人数进行了分析。在此基础上，提出了一个遵循死亡过程的因果模型。即在多元回归模型中构建死亡人数估算模型，第一步将建筑物的结构损伤模式指定为给定地震地震动烈度的结果，第二步通过空间损伤指数计算建筑物的体积损失，最后一步计算被困在倒塌建筑物中的居民的存活率。该模型不仅与传统的简单模型相比，建筑物倒塌对死亡人数的调节更为准确，而且通过引入体积损失指标，也适用于国外各种类型的建筑物。此外，该模型还可以作为事后对策，为搜救活动的最佳分诊决策提供帮助；由于建筑物的倒塌模式对被困人员的生存潜力影响较大，该模型可以提供有关建筑物倒塌模式的信息。少

(1)Research backgroundTo uncover the mechanism of human injury in buildings damaged by devastating earthquakes is indispensable to the fundamental resolution for prevention of the occurrence of deceased. However, conventional studies formulate the death toll estimation directly from the number of building collapses. Such ridiculous situation would be caused by the fact that there has been no contact in the research field between building structural engineers and disaster medical researchers, and they have been advancing the research under the independent organization each other.(2)Objectives of the projectIn order to interdisciplinary discuss the latest results of the research of the earthquake engineering and the disaster medicine. By the application of regression analysis and some statistical work to the database the study secondly tries to propose the death toll estimation model that runs to concrete measures for the human damage reduction. In this project treats the 2004 Barn Earth … More quake in Iran as an earthquake with masonry structure collapse, the 1999 Kocaeri Earthquake in Turkey as an earthquake with RC structure collapse, and the 1995 Kobe Earthquake in Japan as an earthquake with wooden structure collapse.(3)Implementation of the projectKickoff meeting :To discuss the research plan for 3 years with the researchers in the engineering field and the medical field at the first stage, an interdisciplinary symposium for research exchange was executed by the Japan-U.S.cooperation at the Center for Health Science, the University of Texas. The symposium was held by opening to the public, and the state of the symposium was simultaneously dlelivered to the world through the Internet Web.Construction of database :Based on the discussion in the symposium, the definition of building damage was redefined from the viewpoint of death occurrence in seismic damage building for the purpose of database construction for clarifying the death occurrence mechanism. The new definition was visualized as a pattern chart of building damage. Following the definition, we rebuilt the database from the information about dead and damage to building in the Kocaeri Earthquake and the Kobe Earthquake. In addiiton, we consolidated the database by adding the concerning data which various research organizations had collected.Modeling death toll estimation and its application :We elucidated the death causalities by means of event tree analysis for the Kobe Earthquake's database. On the basis of the result, a causal model following the process to death was proposed. That is to say, the death toll estimation model was constructed in multivariate regression model, in which the structural damage pattern of building can be specified as the result for given earthquake ground motion severities at the first step, the volumetric loss of building can be counted by the spatial damage index at the second step, and then the survival rate of inhabitants entrapped in collapsed building can be calculated at the final step. This model is not only accurate in comparison with the conventional simple models, that the collapse of building regulates the number of killed people, but also applicable to various type of buildings in foreign countries by introducing the index of volumetric loss. In addition, this model can help the decision-making for the optimal triage for search and rescue (SAR) activities as a post countermeasure ; because the survival potential of trapped people in collapsed building is affected strongly by the collapsed pattern of building, and the model can provide the information about the collapsed pattern. Less