An interventional study of Locked CTEPH-LLM Diagnostic Decision Support in Chronic Thromboembolic Pulmonary Hypertension (CTEPH), Acute Pulmonary Embolism (PE) and Pulmonary Artery Sarcoma, sponsored by China-Japan Friendship Hospital. Not yet recruiting at 1 site in China. Open to participants aged 18 Years and older. Per ClinicalTrials.gov, last updated 2026-10-29.
Sponsored by China-Japan Friendship Hospital · Not applicable, Interventional, and Diagnostic
This randomized controlled trial aims to evaluate whether a locked large language model (LLM)-based clinical decision support system improves the differential diagnosis of pulmonary artery filling defects among physicians who do not specialize in pulmonary vascular diseases. Participants will evaluate 30 de-identified clinical cases involving chronic thromboembolic pulmonary hypertension (CTEPH), acute pulmonary embolism (PE), and pulmonary artery sarcoma (PAS), with or without LLM assistance. The primary outcome is sensitivity for identifying CTEPH, while secondary outcomes include overall diagnostic accuracy, time to diagnosis, appropriateness of referral recommendations, and AI-related diagnostic errors.
Pulmonary artery filling defects can result from several conditions, including chronic thromboembolic pulmonary hypertension (CTEPH), acute pulmonary embolism (PE), and pulmonary artery sarcoma (PAS). Distinguishing these conditions can be challenging for physicians who do not specialize in pulmonary vascular diseases. This prospective randomized controlled trial aims to assess whether a clinical decision support system using a locked large language model (LLM) can improve diagnostic performance and reduce diagnostic errors. The study will enroll up to 66 physicians, with a target of 60 participants completing the trial. After providing informed consent and completing a baseline assessment, participants will be stratified by professional seniority and randomly assigned in a 1:1 ratio to either an LLM assisted group or a control group. Randomization will be performed using a computer generated sequence, with allocation concealed until baseline assessments are completed. All participants will independently evaluate the same 30 deidentified clinical cases, including 10 cases each of CTEPH, acute PE, and PAS. The diagnoses will be confirmed by relevant specialists using established diagnostic criteria. These cases will not have been used for model development or optimization. Each participant will complete two sessions of 15 cases, separated by 3 to 7 days, with case order randomized for each participant. Clinical information will be presented in four stages: clinical history and physical examination, laboratory findings, echocardiography, and computed tomography pulmonary angiography (CTPA). Participants will record their diagnosis at each stage before receiving additional information. The control group will make diagnostic decisions based on the available clinical information, while the LLM assisted group will also receive the model's most likely diagnosis, supporting and opposing evidence, and referral recommendations. The model version, prompts, parameters, and case specific outputs will be fixed before randomization and remain unchanged throughout the study. The primary outcome is sensitivity for identifying CTEPH after all four stages. Secondary outcomes include diagnostic accuracy across all three conditions, diagnostic performance at each stage, time to diagnosis, appropriateness of referral recommendations, and errors associated with reliance on incorrect LLM recommendations. Analyses will follow the intention to treat principle. Mixed effects logistic regression will be used to compare diagnostic performance between groups, with physician seniority included as a fixed effect and physician and case included as random effects. The study will use deidentified clinical cases and will not involve direct patient participation or influence clinical care.
756 studies on the registry are indexed under Pulmonary Embolism; 159 are open to participants now.
This study's planned enrollment of 60 is below the median of 155 across 390 interventional studies indexed under Pulmonary Embolism.
Browse Pulmonary Embolism studies →China-Japan Friendship Hospital is the lead sponsor of 178 studies on the registry; 111 are open to participants now.
Counted across the registry records on this site, refreshed daily.
Go through the criteria one by one — yes, no or not sure — and take the list to your doctor or the study team. Kept in this browser only.
The study participants are clinicians; patients represented in the de-identified case materials are not enrolled as study participants.
Inclusion Criteria:
Exclusion Criteria:
Clinicians will review the same 30 de-identified cases using the same four-stage sequence and response requirements as the control group. At each stage, they will additionally receive the corresponding locked LLM output, including a Top-1 diagnosis, key supporting and opposing evidence, and referral advice. Clinicians may accept or reject the AI advice and remain responsible for their submitted answers.
Other: Locked CTEPH-LLM Diagnostic Decision Support
Clinicians will review the 30 de-identified cases using the standardized four-stage sequence and will record their Top-1 diagnosis, key diagnostic rationale, referral recommendation, and response time at each stage. They will not have access to the study LLM interface or any other generative AI tool.
A locked LLM-based decision support system will process the same staged clinical information reviewed by participating clinicians. For each case and stage, the system will display a pre-generated Top-1 diagnosis among CTEPH, acute PE, and PAS; key evidence supporting and opposing that diagnosis; and a referral recommendation. Before randomization, the model name and version, call date, system and user prompts, input fields, context organization, sampling parameters, output format, risk thresholds, and all case-stage outputs will be locked. Outputs will not be selected, repeated, or modified on the basis of participant responses during the study.
Stage 4 CTEPH Identification Sensitivity
The proportion of evaluable physician-case judgments for the 10 reference-standard CTEPH cases in which the physician's final Top-1 diagnosis after Stage 4 is CTEPH. The numerator is the number of correct CTEPH identifications and the denominator is the number of evaluable Stage 4 judgments for CTEPH cases. The between-group comparison will account for clustering by physician and case.
Time frame: At completion of Stage 4 for each case, within 14 days after the clinician's first formal case-reading session
Stage 4 Three-Class Top-1 Diagnostic Accuracy
The proportion of evaluable Stage 4 physician-case judgments in which the clinician's Top-1 diagnosis matches the reference diagnosis among CTEPH, acute PE, and PAS.
Time frame: During the two formal case-reading sessions, completed within 14 days after the first session
Stage 4 Class-Specific Diagnostic Performance
For CTEPH, acute PE, and PAS separately, sensitivity, specificity, positive predictive value, and negative predictive value will be calculated from evaluable Stage 4 Top-1 diagnoses against the prespecified reference standard.
Time frame: During the two formal case-reading sessions, completed within 14 days after the first session
Diagnostic Response Time
Total response time across all cases and response time at each information stage will be recorded automatically by the study platform. Time will be summarized in seconds or minutes, using means with standard deviations or medians with interquartile ranges as appropriate.
Time frame: During the two formal case-reading sessions, completed within 14 days after the first session
Stage-Wise Diagnostic Change
Changes in the clinician's Top-1 diagnosis and recorded diagnostic probabilities from Stage 1 through Stage 4 will be summarized, including the proportion of judgments that change between consecutive stages and between Stage 1 and Stage 4.
Time frame: During the two formal case-reading sessions, completed within 14 days after the first session
AI-Assisted Correction Rate When the AI Recommendation Is Correct
Among LLM-assisted judgments in which the clinician's diagnosis before considering the stage-specific AI output is incorrect and the AI Top-1 diagnosis is correct, the proportion in which the clinician submits the correct diagnosis after receiving the AI output.
Time frame: During the two formal case-reading sessions, completed within 14 days after the first session
AI-Related Over-Reliance Rate When the AI Recommendation Is Incorrect
Among LLM-assisted judgments in which the clinician's diagnosis before considering the stage-specific AI output is correct and the AI Top-1 diagnosis is incorrect, the proportion in which the clinician changes to the incorrect AI diagnosis after receiving the AI output.
Time frame: During the two formal case-reading sessions, completed within 14 days after the first session
Plan to share: No
No publications or documents are linked to this record.
Start is estimated for Oct 2026; primary completion for May 2027, 7 months later.
Of 115 non-industry interventional studies with no phase under Pulmonary Embolism, half reached primary completion within 29 months of starting (the median), counting those still running; a quarter within 15 and three quarters within 54. This study's planned 7 months is below that.
Counting every sponsor, half of 140 studies reached it within 26 months; a quarter within 15 and three quarters within 46.
How these figures are worked out, and which studies are left out of them.
From the registry record's own update history. This site started tracking changes on Sep 25, 2026; for anything earlier, see the record history on ClinicalTrials.gov ↗
This study is not yet recruiting, as verified in Jul 2026. You cannot join it, but the record below documents what was studied.
Get an email when the registry record changes — status, dates, results — or when someone posts here.
Sign in to followQuestions and observations about this study, from anyone following it. Not medical advice, and not a channel to the study team — their contact details are on the registry record.
Sign in to join the discussion. Reading takes no account; posting does. You choose a display name, and a pseudonym is the default.
Nothing here yet. If you are running this trial, taking part in it, or weighing whether to, this is the place to say so.
China-Japan Friendship Hospital →