Researchers looked at 44 studies to understand the challenges and helpful strategies for using medical artificial intelligence (AI) in healthcare settings with limited resources. These settings are in low- and middle-income countries. The goal was to map out what problems often come up and what approaches might help overcome them.
The review found that common barriers to using medical AI included unreliable technology infrastructure, problems with collecting and managing health data, limited local familiarity with AI, and a lack of clear rules or governance. On the other hand, studies pointed to enabling strategies. These focused on improving infrastructure, setting data standards, building local skills and capacity, checking AI tools for fairness, and integrating them into existing healthcare systems and rules.
No specific safety concerns from using AI were reported in this review. The main reason to be careful is that the evidence comes from a wide variety of different studies, which could not be combined to give precise, numerical results. Also, the review only included studies published in English, which might miss important experiences from non-English speaking regions.
Readers should take from this that deploying medical AI in low-resource settings is complex, with known hurdles and suggested approaches. The review provides a useful summary of the current discussion and landscape, but the evidence is not yet strong enough to point to one best way forward. It highlights areas that need more focused research and local adaptation.