SWEEP-TB — AI Chest X-Ray Screening Confronts Hidden Tuberculosis on Daru Island, Papua New Guinea

Daru Island, in Papua New Guinea's Western Province, carries one of the highest documented tuberculosis burdens in the world, compounded by drug-resistant strains and limited on-island diagnostic capacity. Since 2019, the Western Provincial Health Authority and PNG's National TB Program, working with the Burnet Institute, the University of Papua New Guinea and the PNG Institute of Medical Research, have run SWEEP-TB, a population-wide active case-finding campaign built around a truck-mounted mobile digital X-ray unit.

The programme set out to screen as much of Daru's population as possible for tuberculosis, using artificial-intelligence-assisted chest X-ray reading to triage suspected cases for confirmatory testing far faster than manual radiologist review would allow on an island with few specialists.

The mobile X-ray unit uses Delft Imaging's CAD4TB AI software to read chest X-rays automatically and triage suspected cases for confirmatory GeneXpert testing. A later phase of the programme ran from June to December 2023, extending screening to a further round of residents.

Of Daru's 18,854 residents, 8,085 (42.9%) were enrolled and 7,970 (98.6%) had their chest X-rays interpreted by the AI software; 69 people were confirmed TB-positive via Xpert testing, a detection rate of 853 per 100,000 screened, published in peer-reviewed research in BMC Public Health. In the June-December 2023 phase, a further 3,263 people were screened, of whom 13.7% (447) screened positive on CAD4TB and symptom criteria and 6.9% (24 of 348 tested) were confirmed TB-positive and started on treatment.

SWEEP-TB is a rare case of AI-assisted screening embedded directly in a government health authority's operational public-health response, rather than a research pilot run in parallel to services. As an island-specific elimination campaign, however, its results have not yet been shown to transfer to PNG's mainland provinces.