Make Time Count Today - Reducing criminal reoffending on probation through data analytics, predictive behaviour recognition and optimised interventions

让时间变得有意义 - 通过数据分析、预测行为识别和优化干预措施减少缓刑期间的刑事再犯罪

• 批准号: 10009205
• 金额: $ 44.1万
• 依托单位: MAKE TIME COUNT TODAY LTD
• 依托单位国家: 英国
• 项目类别: Feasibility Studies
• 财政年份: 2021
• 资助国家: 英国
• 起止时间: 2021 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=10009205
• 关键词: Make; Time; Count; Today; Reducing

**Problem Addressed**Crime costs UK economy over £58bn pa, with 1.2mn people convicted annually. Of these, 87% have previous convictions, 60% of released prisoners and 30% on probation reoffend within 12 months.Reoffending is not inevitable. Evidence from countries such as Norway and Germany have shown that targeted alternatives such as Out of Court Disposal orders _(_OOCD) and active engagement in Initial Sentence Plans (ISPs) is highly effective at reducing these reoffending rates (-20% reoffending rate vs 60%).Despite being part of the Criminal Justice systems arsenal, budget cuts have increased crime rates by 24% and meant that individuals for are 9x likely to be resent to court, rather than be prescribed these more effective alternatives.**Innovation**Utilising AI, Machine Learning (ML) and user-centric ecosystem design, the goal of the system to more frequently engage with user, make them feel supported and prevent downward spirals to reoffending through targeted, proactive inter_v_entions.The project has been devised with and will be delivered in partnership with the Cambridge Centre for Evidenced-Based Policing, part of the University of Cambridge, whose pioneering work on Al, data analytics and Policing has resulted in an algorithm that predict six times as many cases of serious violence as the current UK Government's Offender Assessment System _(_OASys) risk classification model predicted.**Advancement on Current Approaches**Building upon this strong evidence base, we intend to develop a highly-innovative platform to help probation and rehabilitation stakeholders to;1. Identify those individuals on probation that are at the highest risk of reoffending using new machine learning techniques, 6x more accurate than the current OASys system2. Design and initiate tailored, targeted early interventions -- nudges, check-ins, contact time, support sessions -- via our support ecosystem, to prevent downwards spirals towards reoffending.3. Facilitate the safe, secure and easy sharing of user information and ISPs between inter-agencies.4. Reduce the administration and data silos associated with these.Through proactive management, positive nudges and contact time we aim to reduce reoffending by providing timely support and steer to individuals most at risk of reoffending.**_Targeted Impacts: Increase the number of touch points; reduce administration; optimise intervention per individual to reduce the risk of reoffending._**

**问题得到解决**犯罪给英国经济造成的损失超过580亿英镑/年，每年有120万人被定罪。在这些囚犯中，87%有前科，60%获释囚犯和30%正在缓刑的囚犯在12个月内再次犯罪。来自挪威和德国等国的证据表明，有针对性的替代方案，如庭外处置命令_(_OOCD)和积极参与初始判决计划(ISPS)，在降低这些再犯罪率方面非常有效(-20%对60%)。尽管预算削减是刑事司法系统武器库的一部分，但预算削减使犯罪率增加了24%，意味着个人被重新起诉的可能性增加了9倍，而不是被开出这些更有效的替代方案。**创新**利用人工智能、机器学习(ML)和以用户为中心的生态系统设计，该系统的目标是更频繁地与用户互动，让他们感受到支持，并通过有针对性的、主动的干预防止再次犯罪的螺旋式下降。该项目是与剑桥大学下属的剑桥循证警务中心设计并将与其合作实施的，该中心在人工智能、数据分析和警务方面的开创性工作导致了一种算法，该算法预测的严重暴力案件是当前英国政府罪犯评估系统_(_OASYS)风险分类模型预测的六倍。**改进现有方法**基于强大的证据基础，我们打算开发一个高度创新的平台，帮助缓刑和改过自新的利益相关者；1.使用新的机器学习技术识别那些再次犯罪风险最高的缓刑人员，准确率比目前的OASYS系统高6倍2。通过我们的支持生态系统，设计和启动量身定做的、有针对性的早期干预措施--轻推、签到、联系时间、支持会议，以防止再次犯罪的螺旋式下降。促进机构间安全、可靠和轻松地共享用户信息和互联网服务供应商。减少与此相关的管理和数据孤岛。通过积极的管理、积极的推动和联系时间，我们的目标是通过提供及时的支持来减少再犯罪，并引导至最有可能再次犯罪的个人。**_有针对性的影响：增加接触点的数量；减少管理；优化每个人的干预以降低再次犯罪的风险。_**