← Back to all trials
Recruiting
NCT07895290
A Multicenter Randomized Trial of CT-Based AI Decision Support for Early Pneumonia Care
Conditions: Pneumonia; Pulmonary Infection
Sex: All
Ages: 18 Years – N/A
Healthy volunteers: No
Phase: NA
Enrollment: 4800
Sponsor: Union Hospital, Tongji Medical College, Huazhong University of Science and Technology
Location: Union Hospital, Tongji Medical College, Huazhong University of Science and Technology Wuhan Hubei
Summary
Adults hospitalized with pneumonia who need new or increased oxygen support or have defined immunocompromise will be assigned to usual care with or without access to a CT-based AI report. The report summarizes likely pathogen classes, lung involvement, and short-term deterioration risk. Clinicians review the report alongside images and clinical evidence and retain responsibility for testing and treatment. The trial asks whether AI-assisted care reduces death or new invasive respiratory or circulatory support within 28 days. Patients are followed after discharge through day 90.
Eligibility Criteria
Inclusion Criteria:
1. Hospitalized with suspected or confirmed pneumonia and new or progressive infection-compatible chest CT abnormalities.
2. At increased clinical risk: new or increased oxygen requirement above the stable pre-illness baseline attributable to the current pneumonia, OR protocol-defined immunocompromise.
3. A clinically indicated CT meeting the frozen input-quality criteria is available. Randomize within 24 hours after that CT and within 48 hours of documented recognition of the current pneumonia episode.
4. No primary qualifying organ support earlier in the current episode, at baseline, or already clinically indicated before allocation.
5. Written informed consent from the participant or an authorized representative before randomization.
Exclusion Criteria:
1. No interpretable CT suitable for the locked model; a noninfectious diagnosis fully explains the findings without suspected concomitant pneumonia.
2. Qualifying invasive respiratory or circulatory support in the current episode, including support already clinically indicated even if not yet delivered.
3. Previous randomization in this trial; a concurrent interventional study that conflicts with the AI care strategy.
4. Known pregnancy, outside the initial intended-use population; breastfeeding alone is not an exclusion.
5. Exclusive comfort-focused care or a pre-existing ceiling of care that rules out both invasive ventilation and vasopressor support.
6. Consent cannot be obtained under the approved arrangements.
Source: ClinicalTrials.gov (NCT07895290). StuddyBuddy aggregates publicly available trial information.