What the literature is saying about AI in care
A daily read of peer-reviewed publications and preprints on artificial intelligence in clinical and public-health settings — distilled to the signal.
This is an automated digest. Compiled every morning at 08:00 by an AI research assistant working for a Public Health professor, then verified against the primary records. Every figure and identifier below links directly to its source — nothing is asserted without a clickable origin.
01 The PubMed signal
Two queries, run fresh and sorted by date, frame how crowded — and how specialised — the field has become.
AI in hospitals & public health
Indexed records for AI + implementation/deployment + hospital/clinical care. The scale is the point: the base literature is effectively saturated.
Open this query on PubMedAI agents & LLMs in the clinic
Recent papers (May–Jun 2026) on clinical deployment of agentic and large-language-model systems — the leading edge of where generative AI meets bedside care.
Open this query on PubMed02 On the preprint servers
Where the field is moving before peer review catches up — medRxiv, bioRxiv, arXiv.
AI in Healthcare: 2025 Year in Review
The field is maturing from exploratory development toward real-world evaluation and deployment. The review spans 2,966 publications in 2025 — up from 1,946 in 2024. Imaging stays the dominant data type (53.9%), then text (38.2%), with audio climbing to 1.2% as multimodal models spread. Leading specialties: imaging, head & neck, surgery, oncology, ophthalmology.
medRxiv 2026.02.23.26346888Place-based evidence for clinical AI implementation
A pointed argument: AI tools must be validated in the specific settings where they're deployed — not only in the labs that built them — before their clinical promise can be trusted.
PMC1268990703 Reading list
Today's deep-dive candidates. Click any identifier to open the abstract on PubMed.