SUDCare: A smart mobile tool to support substance use screening and follow-up care

项目摘要 物质使用障碍（SUD）是一个重大的公共卫生挑战，每年花费美国超过7400亿美元 并且缺乏现有的劳动力来解决这个问题。我们提出的创新，SUDCare，是一种新颖而智能的 移动的应用程序，旨在解决初级保健患者的迫切需求， 用于SUD筛查和后续护理的资源和服务不足。拟议方案的核心组成部分 创新将包括用于评估患者SUD风险水平的远程筛查工具和聊天机器人 以支持后勤和日常查询。我们的研究团队将利用自然语言处理（NLP）， 机器学习技术，以创建可扩展和可持续的解决方案，以支持基于证据的SUD 筛选和治疗方案。SUDCare有可能简化后续诊所就诊， 最大限度地减少增加的工作量负担，同时填补在照顾难以 覆盖目前没有任何SUD治疗方案直接服务的人群。本SBIR的目标 第一阶段项目是测试SUDCare的技术优势、可行性和商业潜力。要全面完成 这一目标使用以用户为中心的设计框架，遵循三个具体目标。在目标1中，我们将进行半- 与利益相关者进行结构化的形成性访谈（即，初级保健提供者，护士，护理协调员， 保险公司、医疗IT专家）和潜在目标最终用户（即，患者）（子研究1，N=20） 我们的合作初级保健诊所从该子研究中收集的数据将用于支持SUDCare 与临床工作流程和系统集成的开发和要求。在目标2中，我们将建立一个工作 SUDCare的原型。为了创建聊天机器人原型，我们将使用丰富的， 来自先前相关工作的文本消息对话的现有去识别数据集。在目标3中，我们 将通过对存在SUD风险的患者进行的可用性测试评估SUDCare的可接受性 （子研究2，N=8）。第一阶段的结果将为潜在的第二阶段项目提供信息，以进一步发展SUDCare， 开展更大规模的试点研究，以评估SUDCare的效率和有效性。

PROJECT SUMMARY Substance use disorder (SUD) is a significant public health challenge that costs the US over $740 billion annually and lacks the current workforce to address it. Our proposed innovation, SUDCare, is a novel and intelligent mobile application that will be designed to address the pressing needs of primary care patients who have inadequate resources and services for SUD screening and follow-up care. The core components of the proposed innovation will consist of a remote screening tool for the assessment of patients’ risk level of SUD and a chatbot to support logistical and routine queries. Our research team will utilize natural language processing (NLP) and machine learning techniques to create a scalable and sustainable solution to support an evidence-based SUD screening and treatment protocol. SUDCare has the potential to streamline follow-up clinic encounters and minimize the increasing workload burden placed on PCPs while filling a largely unfilled gap in care for hard-to- reach populations not directly served by any SUD therapeutic solutions available today. The goal of this SBIR Phase I project is to test technical merit, feasibility, and commercial potential of SUDCare. We will accomplish this goal using a user-centered design framework following three specific aims. In Aim 1, we will conduct semi- structured formative interviews with stakeholders (i.e., primary care providers, nurses, care coordinators, insurers, health IT specialists) and potential target end users (i.e., patients) (Substudy 1, N=20) recruited from our partnering primary care clinic. The data collected from this Substudy will be used to support the SUDCare development and requirements for integration with clinic workflow and systems. In Aim 2, we will build a working prototype of SUDCare. To create the chatbot prototype, we will train an English language model using a rich, existing de-identified dataset of text-message conversations from previous related work. Finally, in Aim 3, we will assess the acceptability of SUDCare through usability testing conducted with patients at risk of SUD (Substudy 2, N=8). Results from Phase I will inform a potential Phase II project to further develop SUDCare and conduct a larger-scaled pilot study to evaluate the efficiency and effectiveness of SUDCare.