CAD4TB falls just short of WHO's bar for TB screening in Africa

Pooling five African studies of 101,060 people, researchers found that CAD software reading chest X-rays had sensitivity 0.87 and specificity 0.74, just below WHO's target for screening.

  • Not rated an organization

    The review pooled accuracy from screening studies, and it reports no outcome of screening programs that used the software.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Reported and intended benefits.
Evidence

foundstudiedreportedclaimed

An independent peer-reviewed study supports the benefit.

Compared with

WHO's target product profile for a TB screening test. Measured.

Without the system
sensitivity ≥0.90 and specificity ≥0.70
With it
pooled sensitivity of 0.87 and specificity of 0.74
The system’s part
Not isolated. The review measured the software's own accuracy against microbiological tests, with no comparator test such as human readers or symptom screening. The authors describe CAD as AI-based software that scores chest X-rays, and three of the studies set thresholds after analysis.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Not yet checked.

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
CAD4TB, with qXR in one study (computer-aided detection software for TB)
Made by
Delft Imaging (CAD4TB) and Qure.ai (qXR)
Used by
Active case-finding programs and national TB prevalence surveys in Kenya, South Africa and Zambia (five studies)
Where
Kenya, South Africa and Zambia
Stage
In a pilot
Date
, when the source was published
Kind
Aggregate result
Broad benefit
Identifying Objects, Actions, and Events
Last revised
31 October 2026

Sources

  1. Peer-reviewed study, by authors with no tie to the system.

    Scott, A. J., T. Perumal, A. Hohlfeld, et al. Diagnostic Accuracy of Computer-Aided Detection During Active Case Finding for Pulmonary Tuberculosis in Africa: A Systematic Review and Meta-analysis. Open Forum Infectious Diseases 11(2): ofae020, 2024 (published online 17 January 2024).

    A meta-analysis revealed pooled sensitivity of 0.87 (95% CI, 0.78-0.96) and specificity of 0.74 (95% CI, 0.55-0.93), just below the World Health Organization (WHO)-recommended target product profile
    We found a high risk of bias and applicability concerns across all studies.

Details

AI system

Reported configuration

CAD4TB v5, v6 and v7; qXR v3

Workflow

The software scored each chest X-ray for abnormalities that suggest TB, against a threshold. Sputum culture or nucleic acid tests served as the reference standard.

Follow-up

Conditions for replication

The threshold suits the setting, since a more sensitive threshold raises the number of confirmatory tests. Participants are 15 years or older.

Entry details

Sector

Healthcare (tuberculosis screening).

Purpose

Delft Imaging and Qure.ai built CAD4TB and qXR to read chest X-rays and score abnormalities that suggest tuberculosis.

References

Other identifiers

Prospero: CRD42022364968.

Not yet checked: model, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, category mappings.

Cite this record

Peter H. Vartanian. (2026). CAD4TB falls just short of WHO's bar for TB screening in Africa (AIBD-0915). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0915.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions