2026年临床人工智能状况报告

Ethan Goh Adam Rodman Peter Brodeur Jonathan H Chen Dr. Adam Rodman is an assistantprofessoratHarvardMedicalSchool.He is the Director of AIPrograms for the Carl J. ShapiroCenter.Dr.RodmanisanAssociate Editor at NEJM AI. He isalsothe host of the AmericanCollegeof Physicians podcastBedside Rounds. Dr.Peter Brodeur is a risingcardiologyfellowatHarvardMedicalSchool’sBethIsraelDeaconessMedical Center.Dr.Brodeuris an affiliate of ARISE,reviewerfor Nature Medicine&NEJM AI, and former life sciencesstrategyconsultant.His researchfocusesonhumancomputerinteractionandLLMclinicalreasoning. Dr.Ethan Goh is the ExecutiveDirector of ARISE. His research hasbeen featured in The New York Times,The Washington Post, and CNN. Hedirectsthe Stanford Healthcare AILeadershipProgram,and Harvard’sAgentic AI Executive Course. Dr. Gohis a Founding Editorial Board memberand Associate Editor at BMJ DigitalHealth & AI. Dr. Jonathan H Chen is Stanford’sinauguralDirectorforMedicalEducation in AI in the Division ofComputationalMedicine.Hisexpertisecombining human withartificialintelligence to providebetter healthcare than either aloneisfeatured in the popular presswithover 100 publications andawards. ARISE-AI.ORG Message From ARISE Leadership “There are decades where nothing happens; and there are weeks when decadeshappen.”Recent deployments by technology companies, health systems, and regulatorshave made clinical AI more visible and ever more consequential. At the same time, it hasbecome harder to keep up with emerging research. In some areas the literature isfragmented; in others, it simply doesn’t exist yet for the way these tools are being usedtoday. So what actually holds up in practice? The State of Clinical AI Report (2026)was created to look beyond model performancealone to other critical factors that determine real-world impact: how systems areevaluated, how clinicians and AI work together, and where patient risks start to appear. Frontier AI systems are already powerful. What’s needed now is to safely and effectivelytranslate these tools into real-world care. Ethan Goh, Adam Rodman, Jonathan H ChenInvestigators, ARISE Network ARISE-AI.ORG Engagement and Education Free BMIRColloquiaStanfordComputationalMedicine Colloquia Stanford HealthcareAI Leadership &Strategy Program Generative AI andAgentic AI OnlineCourse ●Weekly talks●Thursday 12-1pm PST●Sign up here●Healthcare AI seminars withStanford / industry leaders●Thursday 12 pm PT, free ●Harvard/Stanford faculty,accredited certificate●Summer 2026 ●Application required. CMEand accredited certificate●May 2026 Apply now Get early access Get weekly invites Page 4 The Current Landscape Clinical AI Is Widely Deployed But Poorly Evaluated ●AI is now embedded across health care: 1,200+ FDA-cleared tools and 350,000+ consumer apps havegenerated a $70B market1. Only a minority underwent peer-reviewed evaluation.2 ●Of 691 FDA-cleared AI/ML medical devices (1995–2023), >95% went through the 510(k) clearancepathway, which is predicated on equivalency to existing devices — many of which were approved onsuboptimal evidence.2 ●~50% of FDA device summaries omitted study design, 53% lacked sample size, and <1% reportedpatient outcomes.2 ●95% of device summaries did not report demographic data, and 91% lacked bias assessments, raisingconcerns about safety and equity in real-world use.2 Bridging the gap between adoption and evidence requires supporting clinicians, health system leaders,policymakers, and the public in interpreting available research. ARISE-AI.ORG Top Takeaways 1.Model capability is accelerating, but evidence of real clinical impact remainslimited.Many studies show what models can do in controlled settings; what’sincreasingly needed are prospective studies that show measurable effects on patientoutcomes and care delivery. 2.Frontier LLM models show very uneven performance.They perform extremely wellon complex reasoning tasks, yet break down when uncertainty, missing information, orchanging context is introduced. 3.Clinicians value automation where it reduces administrative and workflow burden,but these use cases remain understudied.Tasks clinicians most want support withare often underrepresented in current benchmarks and evaluations. ARISE-AI.ORG Top Takeaways 4.Patient-facing AI has significant potential to reshape engagement and access, butraises distinct safety concerns.Direct interaction with patients requires much strongerguardrails and scalable oversight systems that do not currently exist. 5.Multimodal clinical AI applications are approaching practical usability.Improvements in base models are enabling applications that integrate unstructured text,images, and other clinical data to support prediction and decision-making in real-worldsettings. 6.FDA clearance is increasing, but near-term clinical adoption will favor narrow,task-specific systems.AI tools that are tightly scoped to specific domains andcontexts are more likely to demonstrate va