A Chatbot Utilizing Machine Learning and Natural Language Processing to Implement the Brief Negotiation Interview to Improve Engagement in Buprenorphine Treatment among Justice-Involved Individuals

项目总结/摘要 最有可能死于阿片类药物过量的人是最不可能得到 救命药参与司法的人从监狱出来的风险最高 过量死亡（比一般人群高8倍），但其中只有1/20 个体接受丁丙诺啡（bup），一种安全有效的药物，已被证明 将过量服用药物致死的风险降低一半。迫切需要促进 这些人中BUP治疗参与的增加。两大障碍 这些人接受BUP是1）系统级障碍，2）个人水平低 动机我们之前的研究表明，提供个人层面的治疗参与 干预措施提高了个人接受bup的比率。因此，我们的解决方案是 通过1）“扰乱” 系统层面的障碍，绕过缓刑制度的部分， 通过使用人工智能， （AI）的聊天机器人进行推荐，以及2）通过以下方式解决个人动机低下的问题： 对聊天机器人进行编程，以提供BNI本身，而不需要训练有素的专业人员。 目标1：设计和开发一个功能原型聊天机器人，以激励BUP参与。 重要性：（a）与所有利益相关者一起进行以人为本的设计（包括焦点小组）;以及（B） 使用ML和NLP创建与移动的应用集成的功能聊天机器人， 应用程序接口服务器和管理员门户。 目标2：对60名被随机分配到BNI聊天机器人的见习生进行为期4周的试点研究 或作为药物治疗（TAU）。 假设1. BNI聊天机器人组将有更高比例的参与者参加 在4周时，他们的第一次bup预约比TAU组（主要结局）。 假设2. BNI聊天机器人组将展示更高的准备程度， 通过尿液毒理学测量，参与BUP治疗的意图和较低的阿片类药物使用 试验，比TAU组在4周（次要结果）。 假设3. BNI聊天机器人组的满意度将高于 TAU小组

PROJECT SUMMARY/ABSTRACT The people at greatest risk of dying from an opioid overdose are the least likely to get life-saving medication. Justice-involved individuals coming out of prison have the highest risk of death by overdose (8x greater than the general population), yet only 1 in 20 of these individuals receive buprenorphine (bup), a safe, effective medication that has been shown to reduce a person’s risk of death by overdose by half. There is an urgent need to facilitate an increase in bup treatment engagement among these individuals. Two of the top barriers to receiving bup for these individuals are 1) system level barriers and, 2) low levels of individual motivation. Our prior research shows that delivering individual level treatment engagement interventions increase the rate at which individuals receive bup. Thus, our solution is to improve engagement in bup treatment among justice-involved individuals by 1) “disrupting” system level barriers by circumventing the pieces of the probation system that are stigmatizing and reduce the chances of a bup referral by using an artificial intelligence (AI)-based chatbot to make the referral, and 2) addressing low individual motivation by programming the chatbot to deliver the BNI itself, without the need for a trained professional. Aim 1: Design and develop a functional prototype chatbot to motivate bup engagement. Milestones: (a) human-centered design (including focus groups) with all stakeholders; and (b) creation of a functional chatbot using ML and NLP that is integrated with a mobile application, an application program interface server, and an administrator portal. Aim 2: Conduct a 4-week pilot study with 60 probationers randomly assigned to BNI Chatbot or Treatment-as-Usual (TAU). Hypothesis 1. The BNI Chatbot group will have a higher percentage of participants attending their first bup appointment than the TAU group at 4 weeks (Primary outcome). Hypothesis 2. The BNI Chatbot group will demonstrate higher ratings of readiness and intention to engage in bup treatment, and lower opioid use, as measured by urine toxicology tests, than the TAU group at 4 weeks (Secondary outcomes). Hypothesis 3. The BNI Chatbot group will demonstrate higher ratings of satisfaction than the TAU group.