Conference: Towards a mass shooting early alert network by modeling 9-1-1 data streams

University Of California-San Diego, La Jolla CA

Investigators

Abstract

Mass shootings are a growing problem in the United States, especially on school campuses. Emergency response to these events is often delayed due to inefficient or overburdened alert and decision-making chains. As public safety events such as mass shootings result in bursts of 9-1-1 call activity in the vicinity of the events, analysis, and modeling of patterns of such calls enabled by recent advances in artificial intelligence and spatiotemporal data mining holds significant promise for a novel intelligent notification network, capable of triggering meaningful alerts even before the information propagates through the legacy response channels. The proposed workshop will bring together experts from the public safety industry, 9-1-1 operators, regulatory bodies, computer science and AI researchers, educators, and policymakers, to jointly explore the opportunities and critical research, technical and organizational challenges, and strategies of using 9-1-1 data streams for early detection of mass casualty events. The workshop will help clarify data needs, concerns, and constraints related to the implementation of predictive analytics tools and services based on 9-1-1 call patterns and lay the groundwork for a prototype early alert system, which could be deployed in multiple jurisdictions across the United States to ensure a more efficient and timely emergency response. Building on the Next Generation 911 (NG911) standards and protocols for multi-channel and multimedia notifications, the predictive analytics models would let public safety professionals transform the nation’s 911 emergency response system, making it possible to trigger and tailor an early response to different emergencies. The workshop discussions will focus on (1) an improved data model for 9-1-1 data that would enable its efficient use in real-time public safety emergency event detection and notification; (2) initial results and challenges of spatiotemporal pattern detection in 9-1-1 and similar massive datasets, particularly from geo-fenced areas such as school perimeters; (3) preliminary results and opportunities for developing predictive models of mass shooting events using state-of-the-art AI and machine learning tools; and (4) user requirements, and technical, organizational and ethical challenges of deploying such models in operational emergency response networks in a robust, interoperable, and trustworthy manner. Achieving these goals, and outlining a comprehensive research agenda to study the critically-important multi-disciplinary problem of efficient response to mass shootings, will require a convergence of diverse cross-sectoral knowledge and expertise that has not been assimilated previously This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.