Predictive Analytics Applied to Integrated Administrative Emergency Response Datasets in Chicago - Resubmission 01

项目摘要 处于行为危机中的个人经常与执法部门和其他急救人员接触。在……里面 研究指出，芝加哥经历精神危机的人的急诊室就诊次数增加了19% 在2009至2012年间，伊利诺伊州任何一个晚上大约14,000名无家可归者中有32% 患有严重的精神疾病。越来越多地，警察往往是第一个对心理和行为健康做出反应的人 危机。不幸的是，这些遭遇更有可能升级为暴力或其他有害的 结果不同于其他紧急情况。作为对两国之间几次高调的负面遭遇的回应 芝加哥警方和精神病患者，包括一名年轻男子被警察枪杀的悲剧 面对行为健康(BH)危机，芝加哥一直致力于提高BH应急响应能力。这座城市是 实施几项旨在缓解BH危机的积极干预措施，包括增加 部署CIT培训的警察，并部署联合机动反应小组。然而，正在进行的挑战是 各机构正在确定哪些紧急呼叫具有BH组件，因此应优先处理 CIT和联合移动团队等量身定做的干预措施。此外，如果没有准确的基线信息，领导者 不能完全理解实施这类举措的影响。本项目寻求获得和 链接参与紧急第一响应的三个城市机构的追溯管理数据集： 芝加哥警察局、芝加哥消防局和应急管理办公室 沟通。建立一个包罗万象的数据集将使研究人员能够研究BH反应趋势， 包括分析高利用率和地理热点，目的是为各机构提供信息，以 主动瞄准稀缺资源并监测进展情况。我们还将就以下问题与利益攸关方进行面谈 改善了这些地点的应急反应。最后，研究小组--与咨询委员会合作 理事会-将开发和评估实时预测工具的业务可行性，以帮助确定 BH调度员的紧急情况。通过整合多个大型管理数据集，此预测工具将 充分利用不同来源的“信号”，最大限度地提高预测精度。这些综合数据的开发 系统和支持工具，以及研究小组和机构利益攸关方之间的持续伙伴关系 承诺为芝加哥的紧急反应带来持久和可推广的改进。vt.给出 芝加哥作为美国第三大城市的地位，这种对我们现有数据基础设施的使用具有 有可能广泛影响其他面临类似公共安全挑战并拥有类似 零散的管理数据结构。因此，我们也寻求广泛传播项目结果。 在整个政策和科学界，包括在领先的科学期刊和 政策制定者和媒体外展。

Project Summary Individuals in behavioral crisis frequently come into contact with law enforcement and other first responders. In Chicago, the study cite, emergency room visits for people experiencing psychiatric crisis increased by 19 percent between 2009 and 2012, and 32 percent of the roughly 14,000 homeless people on any given night in Illinois have a serious mental illness. Increasingly, police are often the first to respond to mental and behavioral health crises. Unfortunately, these encounters have a higher chance of escalating into a violent or otherwise harmful outcome than other emergency situation. In response to several high-profile, negative encounters between Chicago police and persons with mental illness, including the tragic police shooting of a young man experiencing a behavioral health (BH) crises, Chicago has committed to improving BH emergency response. The city is implementing several proactive interventions aimed a deescalating BH crises, including increasing the number of CIT-trained police officers and deploying co-mobile response teams. However, an ongoing challenge for agencies is determining which emergency calls have a BH component and therefore should be prioritized for tailored interventions like CIT and co-mobile teams. Additionally, without accurate baseline information, leaders cannot fully understand the impact of implementing such initiatives. The present project seeks to acquire and link retrospective administrative datasets from the three city agencies involved in emergency first response: Chicago Police Department, Chicago Fire Department and the Office of Emergency Management and Communication. Establishing an encompassing dataset will allow researchers to study BH response trends, including analyzing high utilizers and geographic hot spots, with the goal to equip agencies with information to proactively target scarce resources and monitor progress. We will also conduct stakeholder interviews regarding improved emergency responses at these locations. Finally, the research team – in collaboration with the Advisory Council - will develop and assess the operational feasibility of a real-time predictive tool that would help identify BH emergencies for dispatchers. By incorporating multiple large administrative datasets, this predictive tool will capitalize on diverse sources of “signal,” maximizing prediction accuracy. Development of these integrated data systems and support tools, and ongoing partnership among the research team and agency stakeholders promises to produce durable and generalizable improvements to Chicago's emergency response. Given Chicago's position as America's third-largest city, such uses of our already-existing data infrastructure has the potential to broadly influence other cities that face similar public safety challenges and possess similar fragmented administrative data structures. As such, we also seek to widely disseminate project findings throughout the policy and scientific communities, including publication in leading scientific journals and policymaker and media outreach.