Digital images have quietly become uncalibrated diagnostic tools in primary care. As remote care expands, visual accuracy canβt be assumed. We need pragmatic safeguards before aesthetics continue to trump clinical truth
@oohgpwales
Rotational island GP NHS Orkney (Hoy). Clinical Senior Lecturer Bangor University. Reuben-Clarendon Scholar, Oxford University, working on tele-consultations and what comes next... Chair NICE QSAC. Churchill Fellow 2024-5
Digital images have quietly become uncalibrated diagnostic tools in primary care. As remote care expands, visual accuracy canβt be assumed. We need pragmatic safeguards before aesthetics continue to trump clinical truth
In our open-access paper, we argue this is a patient-safety issue. We call for research, clinician awareness, platform prompts about filters/lighting, better technical standards & potentially a dedicated βhealthcare modeβ in devices
There are no minimum standards for visual fidelity in remote consultations. No clear guidance on when an image is too distorted to rely on. Yet we make triage and referral decisions based on these visuals every day
Generative AI takes this further. Modern filters can remove shadows, adjust skin tone, even regenerate facial regions. The result can be visually convincing but medically misleading. And neither clinician nor patient may know itβs happening
That matters because many diagnoses are colour-dependent. Jaundice, cyanosis, pallor, erythema. Subtle cues of sepsis or anaemia. If software brightens, smooths or βcorrectsβ tone, clinical signs can be suppressed or exaggerated without anyone realising
Smartphones arenβt clinical instruments. Theyβre built to make people look good. Auto white balance, HDR, night mode, compression, AI enhancement & platform processing all alter images. The entire captureβtransmitβdisplay chain can distort colour and detail
General practice now relies heavily on patient-generated photos. Rashes, swollen joints, βdoes this look infected?β, βdo I look a funny colour?β Remote visual assessment is routine. But we rarely stop to ask a basic question: how accurate are the images weβre seeing
New paper alert⦠thread below www.thelancet.com/journals/lan...
Cannon mounted⦠Hoy is ready for Napoleon.
The Hoy Head steaming in to pick us up
At the bus stop bright and early for the next stage of the journey / bus and then ferry over to Hoy. Itβs windy and cold and Iβm grateful for the padded waterproof onesie purchased in the Falklands (yes I did brush my hair this morning, but no amount of brushing can compensate for Orkney wind!)
Made it to Kirkwall. In a rare feat of organisation Iβve bought biscuits IN ADVANCE as thankyous for all the people who always end up going above and beyond to help me as island doctor - volunteer ambulance crew and ferry crew in particularβ¦ impressed by this unusual level of personal organisation π
Switching back from academic to island GP⦠by 10am tomorrow the metamorphosis will be complete. Looking forward to wielding a stethoscope rather than a keyboard and to wearing my many-pocketed-trousers again. (Back in Bangor 9th March, not leaving for good!)
Banner image with screenshot of scientific article from nature Medicine, as well as two panels from the study method and results
β οΈ Despite all the hype, chatbots still make terrible doctors. Out today is the largest user study of language models for medical self-diagnosis. We found that chatbots provide inaccurate and inconsistent answers, and that people are better off using online searches or their own judgment.
You can read the paper here www.nature.com/articles/s41...
Plenty of press coverage tonight for one of the most important papers Iβve been involved withβ¦ www.bbc.co.uk/news/article...
New paper alert! innovations.bmj.com/content/earl...
Lead author and medical student Aleks Tanaka did a brilliant job on this paper looking at whether drones in healthcare are ready to fly (read it to find out!)
Where in Wales? Come and visit if anywhere close to my stomping grounds!
Will share once out but was only accepted this week so might be a while!
Paper coming out soon on exactly this in Lancer primary care
No. 4 #BJGPTop10 Hybrid general practice: Risks of hybrid general practice and its challenges to quality care doi.org/10.3399/BJGP...
@trishgreenhalgh.bsky.social @oohgpwales.bsky.social
#HybridCare #DigitalHealth #HealthcareInnovation #HealthPolicy #HealthTech #PrimaryCare #BJGP #research
Happy new year! If your eyes glaze over when people start talking philosophy in research, then read this! Proud to have been a co-author on this paper explaining why the assumptions we make about the world matter for our research bjgp.org/content/76/7...
Happy new year! If your eyes glaze over when people start talking philosophy in research, then read this! Proud to have been a co-author on this paper explaining why the assumptions we make about the world matter for our research bjgp.org/content/76/7...
Pretty sure Iβve seen it there!
Look what I saw! Itβs missing a mast π
@bjgp.bsky.social any chance of putting qr codes on your research pages to link to the full research paper?
Wow congratulations!
www.theguardian.com/uk-news/2025... the bishop speaks for me⦠the bible is crystal clear on the need to care for the vulnerable and to welcome foreigners. Government policy is out of line here.
βThis is a call to ensure that the future of health is not defined solely by those with the most data, the most money, or the most influence, but by those who centre equity, prevention, and the well-being of populationsβ¦β read more here www.sciencedirect.com/science/arti...
Very proud of son who has passed his driving test, and absolutely horrified at the cost of car insurance for young drivers (and thatβs for a 7 year old corsa!). Any tips?