内容来源：https://research.google/blog/collaborating-on-a-nationwide-randomized-study-of-ai-in-real-world-virtual-care/

内容总结：

谷歌携手医疗集团启动全国性AI临床研究 拟评估对话式AI在真实世界虚拟诊疗中的表现

2026年2月3日，谷歌研究负责人迈克·谢克曼与卡梅伦·陈宣布，谷歌已与美国知名医疗服务提供商Included Health达成合作，计划开展一项全国范围内的前瞻性随机对照研究，以评估对话式人工智能在真实世界虚拟临床工作流程中的实际效用。这项研究目前正等待机构审查委员会批准，预计将成为该领域内首个大规模现实环境下的实证研究项目。

此前，谷歌团队已在模拟环境中通过AMIE等研究系统验证了AI具备临床医生水平的诊断推理与对话能力，并与贝斯以色列女执事医疗中心合作完成了单中心可行性研究，初步验证了系统的安全性。本次全国性研究将在此基础上进一步推进，采用随机对照试验设计，在真实患者群体中比较AI辅助的虚拟诊疗与标准临床实践的效果差异，旨在系统评估AI在多种地域、多种健康条件下的大规模应用潜力。

谷歌强调，这项研究是其多年基础研究成果的延伸。团队在三个关键领域积累了核心技术：一是通过AMIE系统提升诊断与疾病管理推理能力；二是借助个人健康代理研究，探索AI分析可穿戴设备数据、提供个性化健康洞察的潜力；三是开发“导览”式AI助手，帮助用户更高效获取可靠健康信息。这些技术将整合应用于本次研究涉及的AI系统中。

研究团队表示，将AI从实验室模拟环境推向真实临床场景的大规模验证，是确保其安全、有效、负责任地应用于医疗的关键一步。该研究采用与药物临床试验相似的严谨证据生成标准，旨在为AI在医疗领域的规范化应用建立新标杆，最终目标是让高质量医疗资源通过先进AI技术惠及更广泛人群。

据悉，该项目由谷歌研究院、DeepMind、平台与设备部门及健康团队共同推进，是谷歌在医疗AI领域迈向现实应用的重要里程碑。

中文翻译：

在全国范围内开展人工智能在现实虚拟护理中的随机研究
2026年2月3日
研究负责人：迈克·谢克曼、陈卡梅伦

我们与Included Health合作，即将启动一项全国首创性研究，旨在评估对话式人工智能在真实世界虚拟护理工作流程中的表现。这项研究将突破模拟环境与回顾性数据的局限，致力于通过严谨的前瞻性证据，大规模评估人工智能在临床环境中的实际效果。

快速链接

具备临床推理与对话能力的人工智能系统，有望大幅提升医疗专业知识与护理服务的可及性，同时让医生能将更多时间投入到真正关键的病患照护中。然而，负责任地开发这类技术需要严谨的、基于证据的研究方法。过去几年中，我们的团队通过研究系统探索了"技术可能性边界"，在模拟环境中展示了达到临床医生水平的能力。尽管我们已开始在临床环境中测试这些系统的安全性与可行性，但要推进到下一阶段的评估，仍需更严格的规范和更大规模的研究。这需要通过对不同地区、不同病情的更多患者进行对照比较，系统研究人工智能在虚拟护理中的效用与影响。

今天，我们宣布在这条持续探索之路上迈出重要一步：我们已与美国领先的医疗服务机构Included Health达成合作，在获得机构审查委员会（IRB）批准后，将启动一项前瞻性、知情同意的全国随机研究，以评估人工智能在真实虚拟护理环境中的表现。这项新研究将建立在我们既往关于人工智能在诊断管理推理、个性化健康洞察及健康信息导航等领域的基础研究成果之上。

这标志着我们研究进程的重要演进。早期发表于《自然》杂志的研究首次评估了我们人工智能系统的诊断推理能力，包括其对医生的辅助效果。随后，我们在模拟环境中通过专业演员扮演患者，将系统的对话式诊断能力与初级保健医生进行了对比。除了评估能力，我们还探索了以医生为中心的异步监督人工智能模式。我们与贝斯以色列女执事医疗中心合作开展的单中心可行性研究，是对话式人工智能进入真实临床环境测试的第一步。该研究旨在通过安全监督员因安全问题介入的次数等结果指标，验证系统的安全性。初步研究已显示出良好的安全信号，我们期待在完成后公布详细结果。

规模化评估：与Included Health的全国性研究

我们的新研究将超越可行性验证阶段，采用随机对照试验设计，在全国范围招募知情同意参与者。通过在此规模上收集扎实证据，我们旨在更深入地理解相较于标准临床实践，我们的人工智能在真实虚拟护理工作流程中处理患者问诊的实际能力与局限。

这种分阶段研究健康领域对话式人工智能的方法，确保随着研究推进，我们能持续获取关于患者与临床医生体验、人工智能系统安全性与实用性的更多数据，从而负责任地指导后续创新。我们相信，医疗环境中对话式人工智能的负责任发展，应遵循与其他医疗干预措施类似的严格证据生成标准。这是在医疗领域安全部署人工智能、建立患者及护理团队信任的关键一步。

立足严谨研究的坚实基础

这项研究凝聚了谷歌多年来在基础研究领域的积累，我们系统性地探索了构建有用且安全的医疗人工智能所需的核心能力。

诊断与管理推理
我们从AMIE系统攻克医学问诊核心挑战起步。通过患者演员与合成临床场景的研究，我们证明了通过模拟自我博弈训练的人工智能系统，在模拟会诊中的诊断准确性与对话质量可达到甚至超越初级保健医生的水平。我们进一步拓展了这些能力以支持长期疾病管理，使系统能够依据临床指南和患者病史进行推理，规划检查与治疗方案，并能处理多模态证据。

个性化健康洞察
我们认识到健康管理不止于诊室，因此通过个人健康助手（PHA）的回顾性研究，探索了人工智能如何基于个人健康数据进行推理。该研究揭示了多模态模型如何分析可穿戴设备记录的睡眠模式与活动指标，以提供个性化健康指导与洞察。通过协同多智能体架构，我们的PHA展示了人工智能如何集数据科学家、领域专家和健康教练于一身——这些能力对于理解患者的完整健康情境至关重要。这些洞察也为Fitbit实验室的实验提供了参考，例如症状自查工具、医疗记录导航器及护理计划功能，帮助我们理解用户在家评估症状或准备就医时如何获取个性化支持。

健康信息导航
为帮助人们更有效地在线搜索健康信息，我们展示了一种创新的"路径导航"人工智能助手如何通过主动对话引导、目标理解与定制化交流，帮助人们获取更优质的信息。这一系列研究为构建清晰、实用且贴合健康管理实际需求的人工智能交互模式提供了关键见解。

诊断管理推理、个人健康洞察与健康信息导航这三条相辅相成的研究脉络，为我们当前及未来研究所考察的人工智能系统奠定了基石。从实验室验证技术可能性，到真实世界中进行大规模人工智能系统研究，我们正朝着通过前沿医疗智能模型让优质护理惠及每个人的目标迈出关键一步。

结语

我们与Included Health即将共同启动的这项全国随机研究，标志着对话式人工智能在医疗健康领域评估的重要进展。从模拟环境与小规模可行性研究，推进到此次大规模、真实世界的全国随机研究，我们正在为医疗人工智能建立全新的高标准证据生成体系。我们的目标是通过严谨研究，深入理解基于AMIE、PHA及路径导航人工智能等基础研究成果的人工智能系统，如何在虚拟护理工作流程中为真实患者与健康关切提供安全有效的支持。这种分阶段的、基于证据的研究方法，对于确保高质量人工智能医疗护理得以安全、负责任地发展，最终提升全民医疗专业服务的可及性至关重要。

致谢
我们衷心感谢Included Health的合作。这项研究汇集了谷歌多个团队的共同努力，包括谷歌研究院、Google DeepMind、谷歌平台与设备团队以及谷歌健康部门。

英文来源：

Collaborating on a nationwide randomized study of AI in real-world virtual care
February 3, 2026
Mike Schaekermann and Cameron Chen, Research Leads
In partnership with Included Health, we will be launching a first-of-its-kind nationwide study to evaluate conversational AI within real-world virtual care workflows. This research will move beyond simulation and retrospective data and aim to gather rigorous prospective evidence on how AI performs in clinical settings at scale.
Quick links
AI systems capable of clinical reasoning and dialogue have the potential to dramatically increase access to medical expertise and care while giving physicians back time with their patients where it truly matters. However, developing these technologies responsibly requires a rigorous, evidence-based approach. Over the past few years, our teams have explored the "art of the possible" through research systems that demonstrate clinician-level capabilities in simulated settings. While we have begun testing the safety and feasibility of these systems in clinical settings, moving to the next stage of assessing these systems requires additional rigor and scale. It involves studying the utility and impact of AI in virtual care involving more patients across an array of geographies and conditions and with controlled comparisons.
Today, we are announcing a significant step in that ongoing research journey: In partnership with Included Health, a leading US healthcare provider, we will be launching, pending Institutional Review Board (IRB) approval, a prospective consented nationwide randomized study to assess AI in a real-world virtual care setting. This new research will build upon our foundational research on the use of AI for diagnostic and management reasoning, personalized health insights and navigating health information.
This work represents a significant evolution in our research. Early studies published in Nature first assessed our AI system’s diagnostic reasoning capabilities, including its assistive effect for physicians. We then compared the system’s conversational diagnostic capabilities to those of primary care physicians in simulated settings with patient actors. In addition to understanding capabilities, we also explored a physician-centered paradigm with asynchronous oversight of AI. Our initial step toward testing conversational AI in real-world clinical settings was a single-center feasibility study in partnership with Beth Israel Deaconess Medical Center. The study’s goal was to demonstrate the system’s safety based on outcome measures like the number of interruptions by the safety supervisor in response to safety concerns. We have observed strong indications of safety in this initial study and look forward to sharing results when complete.
Evaluation at scale: A nationwide study with Included Health
Our new study will move beyond feasibility to use a randomized controlled trial setup with consented participants recruited nationwide. By gathering robust evidence at this scale, we aim to better understand the capabilities and limitations of our AI for managing patient interactions in real-world virtual care workflows compared to standard clinical practice, for real patients and concerns.
This phased approach to studying conversational AI in health ensures that as the stages of research proceed, more data becomes available about the patient and clinician experience, safety and usefulness of the AI system which will guide subsequent innovation responsibly. We believe that a responsible approach to conversational AI in health settings should adopt high standards of evidence generation, similar to other interventions in medicine. This is a crucial step towards ensuring that AI can be deployed safely in healthcare while building trust with patients and care teams.
Building on our existing foundation of rigorous research
This study is informed by years of foundational research across Google, where we have systematically investigated the capabilities required for a helpful and safe medical AI.
Diagnostic and management reasoning
We began by tackling the core challenge of the medical interview with AMIE. Our research with patient actors and synthetic clinical scenarios demonstrated that an AI system trained with simulated self-play could match or exceed primary care physicians in diagnostic accuracy and conversation quality during simulated consultations. We further advanced these capabilities to support longitudinal disease management, equipping the system to reason over clinical guidelines and patient history to plan investigations and treatments, as well as reasoning through multimodal evidence.
Personalized health insights
Recognizing that health extends beyond the clinic, we also investigated how AI can reason over personal health data through our retrospective research on the Personal Health Agent (PHA). This research explored how multimodal models could analyze sleep patterns and activity metrics from wearables to provide personalized coaching and insights. By using a collaborative multi-agent architecture, our PHA demonstrated how AI can act as a data scientist, domain expert, and health coach all in one, capabilities that are essential for understanding a patient's full health context. These insights also informed experiments in Fitbit Labs, such as the Symptom Checker and Medical Records Navigator and Plan for Care, which help us understand how users access personalized support when assessing symptoms at home and preparing for an upcoming doctor’s visit.
Navigating health information
To support people in their search for health information online, we demonstrated how a novel “wayfinding” AI agent helps people find better information through proactive conversational guidance, goal understanding, and tailored conversations. This stream of research has provided critical insights into how to structure AI interactions that are clear, helpful, and grounded in the practical realities of a health journey.
These distinct threads of research — diagnostic and management reasoning, personal health insights, and navigating health information — have laid the groundwork for the AI system we are examining in this and future studies. By moving from demonstrating the art of the possible in the lab to studying AI systems at scale in the real world, we are taking a critical step toward making high-quality care accessible to everyone through frontier models of medical intelligence.
Conclusion
The upcoming launch of this nationwide randomized study, in partnership with Included Health, will mark a significant step forward in the assessment of our conversational AI in healthcare. By moving from simulated settings and small-scale feasibility studies to this large-scale, real-world nationwide randomized study, we are establishing a new, high standard of evidence generation for medical AI. Our goal is to rigorously understand how AI systems, drawing from foundational research like AMIE, PHA, and Wayfinding AI, can be safe and helpful in virtual care workflows for real patients and concerns. This phased, evidence-based approach is essential for ensuring that high-quality, AI-powered care can be developed safely and responsibly to increase access to medical expertise for everyone.
Acknowledgements
We are grateful for the partnership with Included Health. The study is a joint effort across many teams at Google, including Google Research, Google DeepMind, Google Platforms and Devices, and Google for Health.