Effectiveness of Computer-Aided Detection Chest X-Ray Screening for Improving Tuberculosis Diagnostic Yield in Chinese Primary Health Care Settings: Study Protocol for a Prospective Cluster Randomized Controlled Trial

NCT06963606 | Recruiting DMC

• 2 other identifiers: CAMS&PUMC-IEC-2025-044#24-564
• Study Type: interventional
• Target: 22,000
• Locations: 1 country
• Sites: 1

Brief Summary

The global incidence rate and mortality of tuberculosis (TB) pose a challenge to achieving the goals set out in the tuberculosis eradication strategy and the SDGs by 2030. At present, timely and accessible early detection methods for tuberculosis are still a major obstacle. In this context, the emergence of artificial intelligence (AI), especially the AI-assisted chest X-ray (CXR) in the field of diagnostic imaging, has proved the potential to significantly improve the speed and accuracy of tuberculosis diagnosis. However, the extent to which these technologies can affect the broader tuberculosis care cascade, especially by reducing the diagnostic time in the population level, has not yet been explored. The proposed project plans to use the certified AI-assisted CXR system (JF CXR-1) for tuberculosis screening, which aims not only to integrate AI into the diagnosis process, but also to critically assess its impact on the overall tuberculosis care cascade. The selected location for this project is Yichang City in western Hubei Province, China, which is facing a high TB burden. The city has established a strong city-wide health big data platform ten years ago, providing the basis for this project. The project will first optimize the AI-assisted CXR system through retrospective imaging to validate the accuracy of case screening (Stage Ⅰ). Secondly, the project will shift its focus to the real world, where cluster randomized controlled trials will be conducted in primary-care settings (Stage Ⅱ). In this stage, the effectiveness of the AI-assisted CXR system in reducing the diagnostic time of TB cases will be evaluated by comparing with those settings without using the tool. In stage Ⅲ, the qualitative and quantitative methods will be used to evaluate the generalization, practicality, and feasibility of extending the screening strategy in various community environments. If the AI-assisted screening strategy is proven accurate, effective, and sustainable, it may pave the way for its widespread adoption in primary healthcare institutions and other grassroots areas in China. This can not only improve the timeliness of tuberculosis diagnosis, but also help to allocate medical resources more effectively and significantly reduce tuberculosis-related incidence and mortality, bringing positive changes to global public health. In addition, the results of the project can also provide information for policy decisions and guide the formulation of strategies to prioritize the integration of AI into health care, which can not only fight against tuberculosis but also a series of other diseases.

Conditions

Tuberculosis (TB)

Keywords

tuberculosis; computer-assisted detection

Outcome Measures

Primary Outcomes (1)

• The difference in the diagnostic yield of pulmonary tuberculosis screening in township medical and health institutions between the intervention group and the control group.

  Diagnostic yield = The number of people who visited the hospital for chest DR Examination and were ultimately determined to require further relevant examinations and were diagnosed in the tuberculosis specific disease reporting system/the number of people who underwent chest DR Examination

  pre-intervention (via retrospective analysis of historical data), post-intervention (six and twelve months after the intervention starts)

Secondary Outcomes (2)

• The difference in the average number of days from visiting township medical and health institutions to the diagnosis of pulmonary tuberculosis between patients in the intervention group and the control group

  pre-intervention (via retrospective analysis of historical data), post-intervention (six and twelve months after the intervention starts)

• The differences in the accuracy of different tuberculosis screening strategies between the intervention group and the control group

  pre-intervention (via retrospective analysis of historical data), post-intervention (six and twelve months after the intervention starts)

Study Arms (2)

computer-assisted detection

ACTIVE COMPARATOR

Diagnostic Test: Artificial intelligence-assisted chest X-ray in TB screening

Do not use computer-assisted detection

PLACEBO COMPARATOR

Other: Routine doctors analyze the process of chest X-rays

Interventions

Artificial intelligence-assisted chest X-ray in TB screening DIAGNOSTIC_TEST

After the completion of the chest DR Examination, the chest X-ray was analyzed by the artificial intelligence-assisted system (JF CXR-1) to identify the potential signs of tuberculosis. Meanwhile, the doctor analyzed the results of the chest X-ray. After the analysis results were confirmed, the initial judgment results of the doctor and the analysis results of the artificial intelligence-assisted system were recorded, and the reading results of the artificial intelligence-assisted system were fed back to the doctor. Review the doctor's comprehensive analysis results of the artificial intelligence-assisted system, make a final judgment on the chest X-ray results, and determine whether further relevant examinations (such as etiological examination, CT examination, etc.) are needed. Record the doctor's judgment results. Follow up and record the time of diagnosis reported by the tuberculosis specific disease system.

computer-assisted detection

Routine doctors analyze the process of chest X-rays OTHER

After the chest DR Examination is completed, a regular doctor reviews the films without using an artificial intelligence-assisted system. Once the results of the regular doctor's review are confirmed, a final judgment is made on whether further related examinations (such as etiological tests, CT scans, etc.) are needed, and the doctor's judgment is recorded. Follow up and record the time of diagnosis reported by the tuberculosis specific disease system

Do not use computer-assisted detection

Eligibility Criteria

• Age: 15 Years+
• Sex: all
• Healthy Volunteers: Yes
• Age Groups: Child (0-17), Adult (18-64), Older Adult (65+)

You may qualify if:

• Participants must receive medical treatment at the primary healthcare hospitals in Yichang City, Hubei Province, and underwent chest X-ray examinations. The participants have to meet the following criteria:
• \>15 years old.
• Appearance of tuberculosis-related respiratory symptoms or signs.
• Individuals not previous diagnosed with active pulmonary tuberculosis.
• Capable of completing pathogen examinations and subsequent related inspections.

You may not qualify if:

• Those who meet any of the below criteria will be excluded:
• Diagnosed with extrapulmonary tuberculosis or latent tuberculosis infection during the current visit.
• The quality of Chest X-ray images did not meet the standard requirements.
• Unrecognized identity information participants.
• Withdrawal Criteria
• Participants who are lost to follow-up or who do not complete the follow-up period.
• Participants who experience a sudden and serious illness or choose not to continue participating in the study.

Sponsors & Collaborators

• Yichang Center for Disease Control and Prevention, China: collaborator
• JF Intelligent Healthcare Medical Technology Co: collaborator
• Xuelin Yang: lead
• China Medical Board (CMB): collaborator

Study Sites (1)

Township Health-care settings in Yichang City

Yichang, Hubei, 443000, China

RECRUITING

Related Publications (1)

• Yang X, Zhang H, Jiang W, Xin Y, Dai Z, Li Z, Xiong J, Sun R, Shao J, Yu J, Wang Y, Su X, Liu J, Li Z. Effectiveness of computer-aided detection chest X-ray screening for improving tuberculosis diagnostic yield in Chinese primary healthcare settings: study protocol for a prospective cluster randomised controlled trial. BMJ Open. 2026 Feb 4;16(2):e112124. doi: 10.1136/bmjopen-2025-112124.

  PMID: 41638733 DERIVED

MeSH Terms

Conditions

Tuberculosis

Condition Hierarchy (Ancestors)

Mycobacterium Infections; Actinomycetales Infections; Gram-Positive Bacterial Infections; Bacterial Infections; Bacterial Infections and Mycoses; Infections

Study Officials

• Wang Ye Prof

  chool of Population Medicine and Public Health, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing

  STUDY CHAIR

Central Study Contacts

Yang Xuelin

CONTACT

0536-13964768397; yangxuelin321@163.com

Su Xiaoyou Prof

CONTACT

Study Design

• Study Type: interventional
• Phase: not applicable
• Allocation: RANDOMIZED
• Masking: SINGLE
• Who Masked: PARTICIPANT
• Purpose: SCREENING
• Intervention Model: PARALLEL
• Sponsor Type: OTHER
• Responsible Party: SPONSOR INVESTIGATOR
• PI Title: Project researcher

Study Record Dates

• First Submitted: April 28, 2025
• First Posted: May 9, 2025
• Study Start: September 1, 2025
• Primary Completion (Estimated): December 31, 2027
• Study Completion (Estimated): December 31, 2027
• Last Updated: December 31, 2025

Record last verified: 2025-12

Data Sharing

• IPD Sharing: Will share
• Shared Documents: STUDY PROTOCOL, SAP
• Time Frame: January 2025 - December 2027
• Access Criteria: Personnel who have been approved by the research leader can access the declassified data

Locations

CN (1)