Researchers from the DQBM Krauthammer Lab contributed to a new review in npj Health Systems examining how artificial intelligence is reshaping chest radiography. The article traces the field from conventional machine learning through convolutional neural networks and transformers to multimodal systems combining imaging with clinical text.
The review also examines barriers to clinical deployment, including dataset bias, limited cross-institutional generalisation, explainability and integration into hospital workflows. Looking ahead, the authors highlight multimodal AI and large language models as important directions for more context-aware diagnostic support.
Publication: https://doi.org/10.1038/s44401-026-00087-y