Benson R, Brunsdon C, Rigby J, Corcoran P, Ryan M, Cassidy E, Dodd P, Hennebry D, Arensman E
School of Public Health, College of Medicine and Health, University College Cork, Cork, Ireland.
National Suicide Research Foundation, WHO Collaborating Centre for Surveillance and Research in Suicide Prevention, Cork, Ireland.
Front Digit Health. 2022 Aug 20;4:909294. doi: 10.3389/fdgth.2022.909294. eCollection 2022.
INTRODUCTION/AIM: Data visualisation is key to informing data-driven decision-making, yet this is an underexplored area of suicide surveillance. By way of enhancing a real-time suicide surveillance system model, an interactive dashboard prototype has been developed to facilitate emerging cluster detection, risk profiling and trend observation, as well as to establish a formal data sharing connection with key stakeholders an intuitive interface.
Individual-level demographic and circumstantial data on cases of confirmed suicide and open verdicts meeting the criteria for suicide in County Cork 2008-2017 were analysed to validate the model. The retrospective and prospective space-time scan statistics based on a discrete Poisson model were employed the R software environment using the "" and " packages to conduct the space-time cluster analysis and deliver the mapping and graphic components encompassing the dashboard interface.
Using the best-fit parameters, the retrospective scan statistic returned several emerging non-significant clusters detected during the 10-year period, while the prospective approach demonstrated the predictive ability of the model. The outputs of the investigations are visually displayed using a geographical map of the identified clusters and a timeline of cluster occurrence.
The challenges of designing and implementing visualizations for suspected suicide data are presented through a discussion of the development of the dashboard prototype and the potential it holds for supporting real-time decision-making.
The results demonstrate that integration of a cluster detection approach involving geo-visualisation techniques, space-time scan statistics and predictive modelling would facilitate prospective early detection of emerging clusters, at-risk populations, and locations of concern. The prototype demonstrates real-world applicability as a proactive monitoring tool for timely action in suicide prevention by facilitating informed planning and preparedness to respond to emerging suicide clusters and other concerning trends.
引言/目的：数据可视化是数据驱动决策的关键，但在自杀监测领域，这是一个未被充分探索的领域。为了增强实时自杀监测系统模型，已开发出一个交互式仪表板原型，以促进新出现的集群检测、风险概况分析和趋势观察，并通过直观界面与关键利益相关者建立正式的数据共享连接。
分析了2008 - 2017年科克郡符合自杀标准的确诊自杀病例和开放性裁决病例的个体层面人口统计学和情况数据，以验证该模型。基于离散泊松模型的回顾性和前瞻性时空扫描统计在R软件环境中使用“”和“包进行时空聚类分析，并提供包含仪表板界面的地图和图形组件。
使用最佳拟合参数，回顾性扫描统计返回了在10年期间检测到的几个新出现的非显著集群，而前瞻性方法展示了该模型的预测能力。调查结果通过已识别集群的地理地图和集群发生时间线直观显示。
通过对仪表板原型的开发及其在支持实时决策方面的潜力进行讨论，提出了为疑似自杀数据设计和实施可视化的挑战。
结果表明，将涉及地理可视化技术、时空扫描统计和预测建模的集群检测方法相结合，将有助于前瞻性地早期发现新出现的集群、高危人群和关注地点。该原型展示了其在现实世界中的适用性，作为一种主动监测工具，通过促进明智的规划和准备，以应对新出现的自杀集群和其他相关趋势，及时采取自杀预防行动。
