An AI tool that works elsewhere still needs local validation before going live. NHS guidance says digital technologies "cannot usually be deployed directly 'off the shelf'" and calls it "an essential activity." https://t.co/QtQNaXDmmD
NICE says AI used in evidence generation needs methods that are "robust, transparent and reproducible", and it will publish an AI best practice methods framework by March 2027. "Speed alone is not enough." https://t.co/wOZ2VDo4zu
NICE NG28, type 2 diabetes, no relevant comorbidity: modified-release metformin, then SGLT-2 inhibitor once metformin is at maximum tolerated dose. Metformin contraindicated or not tolerated? SGLT-2 inhibitor alone. https://t.co/u08w5E3KUJ
Choosing a PhD supervisor is about the working relationship, not just the name on the door. Dr Ajantha Abey advises meeting potential supervisors and speaking to their current and previous students. https://t.co/IewLW6LQbM
AI in UK healthcare: under the current framework, the MHRA says scribes that only transcribe or summarise for a clinician to review are not medical devices. Clinicians stay responsible for checking outputs: https://t.co/9OrrBcXIv2
A like or share by someone in pharma can count as proactive dissemination. The PMCPA says activity that "could potentially alert one's connections/followers" is "likely to be considered proactive dissemination." https://t.co/gAHPz7pSjF
Omnichannel HCP engagement under the ABPI Code starts with permission. Clause 15.5 rules out promotional calls, email, texts and other digital communications without prior permission. We go channel by channel: https://t.co/FjHWrTzdoy
NICE has chosen AI as the next NICE Listens public dialogue. In NHS AI Lab research it cites, people "raised concerns about data bias and security" and stressed "human involvement and connection." https://t.co/RyWX6NRlCl
Bespoke or off the shelf software in healthcare? NHS guidance expects commodity products to be considered first, and any decision to build to be clearly documented. We set out the checks both routes must pass: https://t.co/8O4bWtaK4w
Practices in England should check indicator changes, reporting requirements and personalised care adjustments in the QOF guidance for 2026/27. We turn those checks into a practical review approach: https://t.co/jCf3jcg3Nq
@TheLancet@LancetRH_Europe A retrospective cohort: association, not causation. Death HR 4.81 vs 2.03 for MI/stroke/death (57,426 matched). Reasons for stopping weren't recorded. Could stopping mark worsening health? The authors raise this too. (Lancet Reg Health Eur, 9 Nov 2026)
UK industry clinical trial recruitment fell for a fifth year: 17,618 people in 2025/26, down from 19,507, says the ABPI. Its reading: "there is not a lack of willing trial participants within the NHS". https://t.co/ThkOJW2scB
@ShivaniM_KC Agree, worth discussing with anyone who already has vision loss. MHRA and EMA class NAION as a very rare semaglutide side effect (up to 1 in 10,000 people). FDA cites 6.4 vs 3.8 cases per 100,000 person-years, GLP-1 vs comparators. Sources: MHRA, EMA, FDA.
@theheartorg Cardiorenal risk is so often found late. Useful panel. Where do you see the biggest gap in practice: spotting it early, or acting on it once it is spotted?
An ABPI Code advisory board needs a genuine unanswered question, and Clause 24.2 says the need must be documented before the services are requested. We follow the paper trail to the report: #MedicalAffairs https://t.co/NpSvhvkAw6
Treating ADHD in adults in UK practice means reading NICE NG87 alongside the UK SmPCs, the MHRA and NHS England. Our new CMB Communications video for healthcare professionals explores how it applies: https://t.co/oM8QYCw2Pm
Running an AI model locally gives you more control over patient data. It does not remove risks. Answering HL4285 on 4 Nov, the Government said all AI models can present risks "that need to be appropriately managed". https://t.co/o86nC3Issa
Patient data and public chatbots do not mix. Health data is special category data under UK GDPR, and NHS England Digital says a DPIA must be completed before AI is implemented. Where local AI fits: #AIinHealthcare https://t.co/sMx0SPDDUp
MPs call it "unacceptable" that obesity training for healthcare professionals is so limited. A surgeon told the committee that in his experience, medical students got "one lecture in the whole curriculum". https://t.co/953xp7Gd89