Submission ID 131874
| Issue/Objective | The co-epidemic of silicosis and tuberculosis (TB) in South Africa's mining industry affects hundreds of thousands of migrant workers who were recruited to work in South Africa's gold mines - and is compounded by limited access to chest X-ray (CXR) screening, which is one of the clinical findings required to gain access to compensation and social benefits for injured workers and their families. Although artificial intelligence (AI)-based computer-aided detection (CAD) systems for TB have demonstrated impressive accuracy against microbiological standards, validation among silica-exposed populations has been limited. Moreover, well-documented biases hinder CAD utility in diverse patient populations, potentially exacerbating existing healthcare inequities. In this presentation, we describe the challenges in developing CAD systems for TB and silicosis, as examples of how local circumstances can present distinct patterns of disease that may be poorly captured by universal AI models. We argue for adding local health equity as a criterion in judging the value of AI, and for the potential benefits that local public-sector development initiatives could bring. |
|---|---|
| Methodology/Approach | Using a local dataset of 2000 CXRs from silica-exposed Southern African mineworkers, alongside publicly available international datasets and pre-trained CAD models, we present empirical evidence of CAD biases based on analysis of the validated detection of disease. |
| Results | Dimensionality reduction analysis produced visual mappings that demonstrate how local CXRs form a distinct cluster, separate from international images. We also found that, relative to TB, reducing image resolution disproportionately degraded silicosis detection. Further visualizations proved that accuracy metrics alone are insufficient measures of clinical reliability, possibly obscuring deployment failures. |
| Discussion/Conclusion | We conclude that local public-sector CAD development offers a viable alternative to reliance on externally developed systems that likely exclude underserved populations. Addressing CAD deficiencies requires curating population-representative datasets that capture local epidemiology and transparent, open-source development practices that enable peer review and bias correction. Embedding technical and clinical expertise locally can transform AI-based CAD from a potential instrument of digital colonialism into a mechanism that produces contextually appropriate diagnostics while advancing knowledge for equitable AI deployment worldwide. |
| Presenters and Affiliations | Jerry Spiegel University of British Columbia Sean Terespolsky University of the Witwatersrand Annalee Yassi University of British Columbia Richard Klein University of the Witwatersrand Rodney Ehrlich University of Cape Town Warrick Sieve University of the Witwatersrand Joshua Bruton University of the Witwatersrand Hairong Wang University of the Witwatersrand Karen Lockhart University of British Columbia |
