AI for Rare Disease Diagnosis in Real World

A Multicentre, Prospective, Cluster-Randomised, Parallel-Controlled Trial of a Rare-Disease Large Language Model in Real-World Clinical Settings

A multicentre, prospective, cluster-randomised, parallel-controlled real-world effectiveness study evaluating whether a rare-disease diagnostic large language model can improve diagnostic quality, efficiency, and health-economic outcomes for physicians managing patients with suspected rare or diagnostically unresolved disease.

Study Overview

• Status: Not yet recruiting
• Conditions: Rare Diseases; Rare Disorders
• Intervention / Treatment: Other: AI system

Detailed Description

Rare disease patients commonly experience prolonged diagnostic odysseys rooted in limited rare disease recognition, phenotypic heterogeneity, and dispersed diagnostic clues. Diagnostic decision-support large language models may improve first-visit consultations by integrating prior records, generating structured analyses, and proposing candidate diagnoses, thereby shortening diagnostic pathways and improving appropriate genetic testing referral.

Here, rare diseases clinics are established at 12 clinical institutions using a two-level randomisation structure: (1) cluster randomisation of physicians stratified by seniority tier and centre, with physician arm fixed throughout to eliminate contamination; (2) eligible consented patients randomly assigned to consultation rooms, blinding patients to arm assignment and providing allocation concealment.

In the intervention arm, patients interact with the AI before their consultation; physicians will review the AI report and interact with AI when seeing patients. In the control arm, patients are seen under standard hospital workflow without any generative AI tools. Outcomes adjudicated by an independent Expert Committee blinded to arm assignment; adjudicators access no AI-generated materials.

• Study Type: Interventional
• Enrollment (Estimated): 1055
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Shuyang Zhang, MD, PhD; Phone Number: +86-13911667211; Email: shuyangzhang103@163.com

Study Locations

• China
  • Beijing, China
    • Peking Union Medical College Hospital
    • Contact: Shuyang Zhang; Phone Number: +86-13911667211; Email: shuyangzhang103@163.com
  • Cangzhou, China
    • Cangzhou Central Hospital
    • Contact: Yong Li; Phone Number: +86-0317-2075013; Email: czszxyyirb@163.com
  • Changchun, China
    • Changchun Sacred Heart Hospital
  • Dongguan, China
    • Dongguan People's Hospital
  • Foshan, China
    • First People's Hospital of Foshan
  • Guiyang, China
    • Guizhou Provincial People's Hospital
  • Jilin City, China
    • Jilin Central General Hospital
  • Kunming, China
    • The First People's Hospital of Yunnan Province
  • Lhasa, China
    • Tibet Autonomous Region People's Hospital
  • Tianjin, China
    • Tianjin Children's Hospital
  • Wuhai, China
    • Wuhai People's Hospital
  • Xining, China
    • Qinghai Provincial People's Hospital
  • Zhangzhou, China
    • Zhangzhou Municipal Hospital of Fujian Province

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: No

Description

Patient Inclusion Criteria:

• Any age. Legal guardian co-signs consent for minors or individuals lacking legal capacity.
• Diagnostically unresolved or suspected rare disease, with at least one prior complete clinical evaluation at a secondary-level or higher institution yielding no confirmed explanatory diagnosis.
• First presentation to the enrolling institution for the current condition, with no prior records in the institutional HIS or outpatient system.
• No prior genetic testing related to the current condition; no results or reports available.
• Written informed consent provided voluntarily by patient or legal guardian, with commitment and ability to complete structured follow-up.

Patient Exclusion Criteria:

• Confirmed diagnosis (clinical, pathological, or molecular) explaining the primary symptoms.
• Emergency presentation, critical illness, or any condition incompatible with trial participation.
• Neither patient nor legally authorised proxy able to complete follow-up.
• Concurrent enrollment in another interventional study with diagnostic accuracy or genetic testing yield as a primary endpoint.
• Prior use of another AI system has already yielded a confirmed diagnosis for the current condition.

Physician Inclusion Criteria

• Licensed physician in internal medicine, neurology, pediatrics, general medicine, rare disease, or a related specialty.
• ≥2 years of clinical practice; competent to manage rare disease patients; stratified into junior or senior tier.
• Voluntary participation with written informed consent.

Physician Exclusion Criteria

• No longer in clinical practice, or unable to fulfill required outpatient duties during the study period.
• Unwilling to provide informed consent or to permit protocol-required collection of consultation and questionnaire data.
• Currently enrolled in another AI-assisted clinical workflow, or expected to be unable to comply with the procedures.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Diagnostic
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Single

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: AI system; Patients and physicians will use the study AI system prior to and during the encounter in addition to conventional clinical workflow. Use of other generative AI tools is prohibited.	Other: AI system; Prior to consultation, the AI system will interact with patients to build a personalised medical profile, generate a structured clinical analysis, and recommend candidate diagnoses. During the encounter, the AI system will interact with and be reviewed by physicians.
No Intervention: Standard of care; Patients proceed directly to the consultation room without AI interaction. The attending physician conducts the encounter per standard hospital workflow using conventional clinical resources only. Use of any generative AI tool is prohibited.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Top-3 Candidate Diagnostic Accuracy	The proportion of patients whose physician-provided up to 3 candidate diagnoses at the first study visit include at least one concordant with the final reference diagnosis (denominator: all randomised patients with a blinded-adjudicated confirmed reference diagnosis; intention-to-treat population).	From the first visit to final reference diagnosis adjudication, an average of 8 weeks.
Rare Genetic Disease Diagnostic Yield	The proportion of all randomised patients who underwent genetic testing and had a clinically significant genetic variant (pathogenic or likely pathogenic) confirmed by an independent review committee blinded to arm assignment (denominator: all randomised patients). Variant pathogenicity classified per predefined ACMG/AMP guidelines.	From the first visit to final reference diagnosis adjudication, an average of 8 weeks.

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Overall Diagnostic Yield	Proportion of all randomised patients who receive a confirmed clinical diagnosis (confirmed by independent expert review) during the trial period.	From the first visit to final reference diagnosis adjudication, an average of 8 weeks.
Overall Diagnostic Accuracy	Among patients with a confirmed clinical diagnosis, the proportion whose diagnosis is concordant with the blinded-adjudicated final reference diagnosis.	From the first visit to final reference diagnosis adjudication, an average of 8 weeks.
Time to Diagnosis	Days from first visit to obtaining clinical diagnosis within the follow-up period. Patients without a confirmed diagnosis censored at end of follow-up.	From enrollment to the end of follow-up, up to 8 weeks.
Consultation Duration	In-room consultation time will be recorded, measured, and compared between arms. Pre-consultation AI interaction time and total visit time are additionally recorded for the intervention arm to fully characterise efficiency impact.	Assessed at each consultation (day 1), within 1 day.
Cost to Diagnosis	Cumulative costs from first visit to confirmed diagnosis (or end of follow-up for undiagnosed patients), including direct medical costs, direct non-medical costs, and indirect costs.	From enrollment to the end of follow-up, up to 8 weeks.
Physician-Reported Experience	Physicians will assess their experience of the diagnostic workflow. Responses will be recorded using a standardized rating scale (range 1-5, where higher scores indicate more positive experience).	Assessed at each consultation (day 1), within 1 day.
Patient-Reported Experience	Patients will assess their experience of the diagnostic workflow. Responses will be recorded using a standardized rating scale (range 1-5, where higher scores indicate more positive experience).	Assessed at each consultation (day 1), within 1 day.
Genetic Testing Referral Rate	Proportion of all randomised patients for whom the physician autonomously decides to refer for genetic testing, capturing the effect of AI intervention on referral behaviour.	Assessed at each consultation (day 1), within 1 day.
Positive Rate Among Referred Patients	The proportion in whom a clinically significant (pathogenic or likely pathogenic) genetic variant was detected among patients who completed genetic testing.	From the first visit to final reference diagnosis adjudication, an average of 8 weeks.
Cost-Effectiveness Analysis	Incremental cost-effectiveness ratio (incremental cost per additional correct diagnosis) as primary metric combining diagnostic yield with cumulative medical costs.	From the first visit to diagnosis or the end of follow-up, up to 8 weeks.

Collaborators and Investigators

• Sponsor: Peking Union Medical College Hospital
• Collaborators: Guizhou Provincial People's Hospital; First People's Hospital of Foshan; Dongguan People's Hospital; Qinghai People's Hospital; Cangzhou Central Hospital; The First People's Hospital of Yunnan; Tibet Autonomous Region People's Hospital; Tianjin Children's Hospital; Zhangzhou Municipal Hospital
• Investigators: Principal Investigator: Shuyang Zhang, MD, PhD, Peking Union Medical College Hospital

Study record dates

Study Major Dates

• Study Start (Estimated): June 20, 2026
• Primary Completion (Estimated): July 1, 2027
• Study Completion (Estimated): December 1, 2027

Study Registration Dates

• First Submitted: June 7, 2026
• First Submitted That Met QC Criteria: June 14, 2026
• First Posted (Actual): June 16, 2026

Study Record Updates

• Last Update Posted (Actual): June 16, 2026
• Last Update Submitted That Met QC Criteria: June 14, 2026
• Last Verified: June 1, 2026

More Information

Terms related to this study

• Keywords: diagnosis; cost-effectiveness; AI; rare diseases; LLM
• Additional Relevant MeSH Terms: Pathologic Processes; Disease Attributes; Pathological Conditions, Signs and Symptoms; Disease; Rare Diseases; Artificial Intelligence; Algorithms; Mathematical Concepts; Intelligent Systems
• Other Study ID Numbers: PUMCH I-26PJ0002

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No