A new review of 29 sources looks at whether AI tools actually help people get the care they need. The strongest evidence shows AI can reduce digital access barriers, but only when paired with careful oversight and validation for different patient groups. The review also found that subgroup underperformance and governance concerns are real issues that need attention.
The review looked at AI-enabled digital health, telehealth infrastructure, predictive analytics, clinical decision support, generative AI, and AI-assisted screening. It found that AI-assisted screening and digitally mediated remote support can offer conditional benefits in targeted settings when high-touch support is built into implementation. However, evidence on generative AI and language-support tools is comparatively smaller, newer, and concentrated in representational-bias applications, with limited evidence of sustained downstream equity gains.
The review notes that equity gains are conditional rather than automatic. It also highlights that effects remain difficult to interpret because different AI tools operate through different equity mechanisms. The evidence depth varied substantially across modalities, and the review distills recurring patterns into a synthesis-derived operational roadmap for procurement, validation, implementation, and post-deployment monitoring.
