OMS
Large multimodal models can combine text, image, audio and other signals, but they require governance proportional to the clinical and institutional risk.
WHO guidance on ethics and governance of AI for health, large multimodal models, risks, evidence, transparency and supervision.
The WHO's 2024 guidance on large multimodal models in healthcare emphasizes ethics, governance, validation, transparency, accountability and care with uses in patient care, research and public health.
The value of the model depends on evidence, safety, human supervision, fit to the local context, data protection and monitoring after deployment.
Multimodal models can support documentation, search, image analysis and knowledge synthesis. In a clinical setting, each use needs its own purpose, limit and validation.
