Just your friendly reminder that #GPT4O terms of use prohibit use for medical purposes, and that medical device regulations apply for any intended purposes such as diagnosis, monitoring, treatment, alleviation or prevention of a disease or injury.

Carry on.

Looking to interview people and companies trying to implement AI in the NHS - get in touch via DM if that's you, whether internal or external

Brilliant stuff from Xiao Liu and Alastair Denniston

Spot on summary of the hype around medical AI, the scientific and regulatory realities, and the need to remain true to an evidence-based approach.

Do EPIC's generative AI replies to patient queries actually save time?

No.

17.9% INCREASED reply length
21.8% INCREASED read time
-5.9% DECREASED reply time (non-significant)

...'not anywhere near where it needs to be to be useful'

jamanetwork.com/journals/jaman…

Do you know what would be even better?

Providing the required documentation for developers to build *regulated* applications off your model.

Training data details, risk management processes, QA documentation etc

Otherwise, you can't use it 'anywhere without restrictions'.

NEW PAPER out today in @BMJ_latest TRIPOD+AI: reporting recommendations for studies developing or validating prediction models for use in healthcare that use #machinelearning methods

bmj.com/content/385/bm…

#ArtificialIntelligence #AIstandards #OpenAccess

Please share 🙏

Me saying stuff to Dr. James Somauroo for an hour.

Have a listen.

youtube.com/watch?v=ffOig7…

Clinical AI summarisers - take note.

LLMs are not particularly good at summarising facts, or including all relevant details.

Medic? Biomedical engineer? Software developer?

Thought about a career in quality assurance for cutting edge medical technology?

Come join us.

Gaining U.S. FDA approval is a challenging process.

However, Envisionit Deep AI® were able to leverage our close understanding of the regulatory landscape and their proprietary validation platform, to overcome the challenges and secure 501(k) clearance.

Listen to their CEO explain⬇️

An encyclopaedia that sometimes just makes stuff up just plausibly enough for you not to notice.

Peer review is broken.