Study of a PST-Trained Voice-Enabled Artificial Intelligence Counselor (SPEAC) for Adults with Emotional Distress

针对患有情绪困扰的成年人的经过 PST 培训的语音人工智能咨询师 (SPEAC) 的研究

• 批准号: 10671735
• 负责人: Olusola A. Ajilore
• 金额: $ 108.83万
• 依托单位: UNIVERSITY OF ILLINOIS AT CHICAGO
• 依托单位国家: 美国
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-07-03 至 2025-06-30
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10671735
• 关键词: Address; Adult; Affect; Amygdaloid structure; Anxiety; Area; Artificial Intelligence; Attitude; Brain; Clinical; Cognition; Cognitive; Computer software; Development; Devices; Diagnosis; Dorsal; Dose; Eligibility Determination; Emotional; Emotions; Engineering; Enrollment; Evaluation; Focus Groups; Functional Magnetic Resonance Imaging; Functional disorder; Future; Generalized Anxiety Disorder; Health; Health Promotion; Health Resources; Human; Image Analysis; Intelligence; Lateral; Learning; Length; Machine Learning; Measurement; Measures; Mediating; Mental Depression; Mental Health; Mental Health Services; Moods; Names; Neurosciences; Outcome; Participant; Patient Outcomes Assessments; Patient Self-Report; Patients; Personal Satisfaction; Persons; Phase; Prefrontal Cortex; Problem Solving; Procedures; Process Measure; Productivity; Professional counselor; Protocols documentation; Psychotherapy; Public Health; Quality of life; Questionnaires; Randomized; Research; Secure; Specific qualifier value; Stress; Surveys; Symptoms; Techniques; Testing; Therapeutic; Time; Training; Translating; Treatment Efficacy; United States National Institutes of Health; Voice; Voice Training; Waiting Lists; Walking; active control; active method; anxiety symptoms; arm; behavioral health intervention; clinical decision support; cognitive control; connected care; coping; cost; depressive symptoms; digital medicine; disability; efficacy testing; emotional distress; experience; follow up assessment; formative assessment; improved; innovation; insight; intelligent agent; iterative design; knowledge base; meetings; neural; neural circuit; neuromechanism; new technology; novel; participant enrollment; pilot test; problem solving therapy; prototype; randomized, clinical trials; recruit; social stigma; targeted treatment; theories; treatment arm; treatment optimization; usability; user centered design; virtual; virtual health

PROJECT SUMMARY BACKGROUND: Depression and anxiety are the leading causes of disability and lost productivity, and are often underdiagnosed and undertreated owing to access, cost, and stigma barriers. Novel and scalable psychotherapies are urgently needed. Advances in artificial intelligence (AI) offer a transformative opportunity to develop intelligent voice assistants as virtual health agents accessible on personal devices. Meanwhile, major advances in human neuroscience have fueled a paradigm shift to study brain mechanisms underlying behavioral health interventions. OBJECTIVES: Leveraging our collaborative team’s transdisciplinary expertise in these emerging areas, we will develop and rigorously test a novel voice-enabled, AI virtual agent named Lumen, trained on Problem Solving Therapy (PST), for patients with moderate, untreated depressive and/or anxiety symptoms. We will investigate the effect of Lumen on engagement of a priori neural targets—amygdala for emotional reactivity and dorsal lateral prefrontal cortex (DLPFC) for cognitive control—as putative mechanisms. DESIGN/ METHODS: The project has 2 phases. In the R61 phase (years 1-2), we will further develop Lumen building on the current prototype and conduct iterative user-centered design evaluations that include focus groups, scenario-based clinician evaluations, and a formative user study with 20 participants. We will pilot test Lumen in a 2-arm randomized clinical trial (RCT, Study 1), with 60 participants with depression and/or anxiety randomized in a 2:1 ratio to receive PST with Lumen (n=40) on a secure study iPad or be on a waitlist (n=20). At weeks 0 and 14, participants will complete functional magnetic resonance imaging (fMRI) to assess neural target engagement as well as validated surveys of patient-reported outcomes (e.g., depressive and anxiety symptoms, functioning, quality of life). In addition, they will complete naturalistic end-of-day assessments of mood, stress, appraisal and coping for 7 days every 2 weeks. If the Go milestone criteria are met, the R33 phase (years 3-5) will include a 3-arm RCT (Study 2) with 200 new participants randomized in a 2:1:1 ratio to 1 of 3 arms: Lumen (n=100), waitlist control (n=50), and in-person PST as active control (n=50). Participants will complete baseline and follow-up assessments using a refined measurement protocol based on Study 1. SPECIFIC AIMS: R61 aims are to (1) establish the functionality, usability, and treatment fidelity of Lumen; and (2) demonstrate feasibility, acceptability, and neural target engagement according to pre-specified Go milestone criteria. R33 aims are to (1) confirm neural target engagement by a superiority test (primary) comparing the Lumen and waitlist control arms and a noninferiority test (secondary) comparing the Lumen and in-person PST arms; and (2) examine the relationships of target engagement to outcomes. The results will provide the basis for future confirmatory efficacy testing of Lumen. IMPACT: This project’s public health impact lies in that a mechanistically tested, PST-trained AI agent could bring proven psychotherapy to people with depression/anxiety who do not seek professional help or who desire more personalized, connected care.

项目概要 背景：抑郁和焦虑是导致残疾和生产力下降的主要原因，并且 由于准入、费用和耻辱障碍，常常得不到充分诊断和治疗。新颖且可扩展 迫切需要心理治疗。人工智能 (AI) 的进步提供了变革机遇 开发智能语音助手作为可在个人设备上访问的虚拟健康代理。同时， 人类神经科学的重大进展推动了研究大脑机制的范式转变 行为健康干预措施。目标：利用我们协作团队的跨学科专业知识 在这些新兴领域，我们将开发并严格测试一种新颖的、支持语音的人工智能虚拟代理，名为 Lumen 接受过问题解决疗法 (PST) 培训，针对患有中度、未经治疗的抑郁症和/或抑郁症的患者 焦虑症状。我们将研究 Lumen 对先验神经目标——杏仁核参与的影响 用于情绪反应和背外侧前额叶皮层 (DLPFC) 用于认知控制——假设 机制。设计/方法：该项目分为两个阶段。在R61阶段（第1-2年），我们将进一步 在当前原型的基础上开发 Lumen 并进行以用户为中心的迭代设计评估 包括焦点小组、基于场景的临床医生评估以及 20 名参与者的形成性用户研究。我们 将在一项 2 臂随机临床试验（RCT，研究 1）中对 Lumen 进行试点测试，该试验有 60 名患有抑郁症的参与者 和/或焦虑症以 2:1 的比例随机分配，在安全研究 iPad 上或在安全研究 iPad 上接受 Lumen (n=40) PST 候补名单（n=20）。在第 0 周和第 14 周，参与者将完成功能磁共振成像 (fMRI)，以 评估神经目标参与度以及对患者报告的结果进行验证的调查（例如抑郁症） 和焦虑症状、功能、生活质量）。此外，他们还将完成自然主义的一天结束 每两周进行 7 天的情绪、压力、评估和应对评估。如果 Go 里程碑标准是 R33 阶段（第 3-5 年）将包括一项 3 臂随机对照试验（研究 2），其中 200 名新参与者随机分组 与 3 个组中的 1 个的比例为 2:1:1：Lumen (n=100)、候补名单控制 (n=50) 和亲自 PST 作为主动控制 (n=50)。 参与者将使用基于以下内容的精细测量协议来完成基线和后续评估 研究 1. 具体目标：R61 的目标是 (1) 建立 流明; (2) 根据预先指定的内容证明可行性、可接受性和神经目标参与 达到里程碑标准。 R33 的目标是 (1) 通过优越性测试确认神经目标参与（主要） 比较 Lumen 和候补名单对照组以及比较 Lumen 和候补名单的非劣效性检验（次要） 亲自进行 PST 武器； (2) 检查目标参与度与结果的关系。结果将 为今后Lumen的验证性功效测试提供基础。影响：该项目对公共卫生的影响 关键在于，经过机械测试、经过 PST 训练的人工智能代理可以为患有此症的人带来经过验证的心理治疗 不寻求专业帮助或需要更个性化、互联护理的抑郁/焦虑人群。