
Hello healthtech fans. This week⊠Googleâs chatbot gets into The Lancet, a tiny sensor gets big money, and still nobody knows what happens when AI stacks it in healthcare.
News Bites đ„Ș
đ©ș Googleâs AMIE Gets Its Lancet Moment (With Doctors Watching Over Its Shoulder): I might be missing something here, but 98 patients chatted to a bot called AMIE before their urgent care visits while clinicians watched the conversations live. None had to be stopped on safety grounds, the AI diagnosis matched the final one most of the time, and clinicians said the summary helped them prepare in 75% of cases. One of the results mentioned was âenhancing the clinician-patient relationship,â though Iâm not sure how that was measured. Also, itâs a feasibility study, not outcomes, and n=98. Head scratcher how this is in the Lancet if Iâm honest, but it looks like the start of Google running purely-AI diagnostic pathways in parallel to what happens now with those pesky humans. If I had to guess, âhelping clinicians prepareâ and âenhances relationshipsâ creates a bit of justification along the way. If I wanted to plot the path to AI-only pathways, this is how Iâd do it. Purely my opinion.
â ïž Four Deaths Linked to Philipsâ Point-of-Care Ultrasound: This is bad. And exactly what we (in healthtech) all need to be aware of to put principles in place so this NEVER happens. Philipsâ Lumify handheld ultrasound caused 4 deaths and 8 serious injuries and the fault was in the software.

See above image so I donât have to explain the tech beyond âsmartphone + ultrasound probe.â Basically, the app can suddenly fail to recognise the probe and block scanning until the probe is registered again, but⊠that needs internet. If you donât have it, youâre not using the ultrasound. Now, why would you have this device? Probably because youâre somewhere like a helicopter, an ambulance or the side of the road with someone critical. When do you need the result? Now. Imagine turning to the screen and seeing âPlease register deviceâ and âNo wifiâŠâ I feel for the patients, their families and the clinicians affected by this. Worth noting that I donât know the exact circumstances of the injuries or deaths, but letâs all agree we must do better. The FDA has issued an alert.
đ Healthleap raises $38M for AI that flags patient decline: Weâve heard this a fair amount over the years, but at the current pace of technology improvement and general regulatory awareness and alignment, Iâm getting more optimistic. This one reads clinical notes, labs and vitals every night and puts risk scores into the care teamâs workflow by morning. Starts with malnutrition and delirium and theyâve raised $38m across seed and Series A, with Sequoia and First Round in at seed and Hummingbird leading the $30m A. Itâs live in 50+ hospitals, (only three a year ago which shows the pace of adoption) and it flags rather than diagnoses, which, I imagine, keeps the regulatory burden manageable.
𩞠Sava bags ÂŁ27m to prove a 10x smaller CGM: I actually had RafaĂ«l on the podcast a while ago and visited their lab. Glad to see theyâve done this Series B because theyâre seriously impressive. Theyâve made a sensor that sits just beneath the skin and itâs about 10x shorter than a conventional CGM filament. The round was led by Ascensia, PHC Holdingsâ diabetes arm, with Balderton, Norrsken VC, Simplyhealth Ventures, JamJar and Pentland also in. If youâre interested in more, have a listen to the podcast.
And finallyâŠ
đ Do GLP-1s cause nail detachment, hair loss, and scalp inflammation? 575 GLP-1 users were compared with 1,329 people with type 2 diabetes and/or obesity. Here are the numbersâŠ
nail detachment in 39% vs 9%
discolouration in 46% versus 17%
unusual hair loss in 50% versus 34%
Iâm now in the habit of just listing the caveats with studies in here because anyone who clicks just wants those, so⊠itâs cross-sectional, self-reported, no standardised exam, half of users had nail problems before starting, and a dermatology brand âsupportedâ the work. So⊠association, not cause. And conference presentation not peer-reviewed paper. Still, I doubt anyoneâs put those in the leaflet.

What to Listen to đ
Tracking, tracking and more tracking. Science Weekly ask if CGMs could benefit those without diabetes. Are we seeing CGMs fall into a wellness accessory now? If so, how would that make those feel who have to wear them to protect their life? Not quite the same, but when people started wearing fake glasses (as a lifelong glasses wearer), Iâd have traded for 20/20 vision in a heartbeat.
If thereâs something that rivals the tracking epidemic, its the GLP-1 discourse. This take is from the Observerâs science editor, whoâs on them himself, which seems more like an investigative exercise. James tracked down the scientist who discovered the hormone behind them, and now has his own concerns about how theyâre being used (the phrase, no sh*t Sherlock, comes to mind).
Homework before you read the governmentâs reply (sorry). Last month HSJâs tech team sat down with Annabelle to pick through the AI Commissionâs final report, the one that promised the NHS a new regulatory rulebook. A tougher MHRA and actual fines for AI firms that harm patients were on the table. Listen with this weekâs response open in another tab and you can tick off what made it through. My bet is youâll get to liability and find the page still blank.
Events đ
â Platform Approaches to Regional AI Implementation and Monitoring
đ
Monday 19 October, 12pm to 1pm
đ Online
Our featured event happening next week is from the Newtonâs Tree team who help hospitals deploy, monitor and govern clinical AI at scale. Theyâre hosting a lunchtime panel on how NHS organisations can roll out AI and keep it in check once itâs live. And itâs epic timing⊠AI diagnostics funding (AIDF round 2) has opened and the MHRAâs National Commission has just published its recommendations on regulating AI in healthcare. CEO Dr Haris Shuaib is joined by the MHRAâs Director of Strategy, Dr Ed Middleton, Prof David Lowe from CERSI-AI and Dr Claire Morris from Cambridge University Hospitals, so youâll hear from the regulator, the people putting regulation into practice and a trust doing the adoption. If youâre buying, building or governing AI in the NHS, this is an hour extremely well spent.
Harley Street Health District Conference
đïž 20 Oct
đ đŹđ§ 1 Wimpole Street, London.
Lord Kakkar opens the line-up and Matthew Syed closes the day, with investors, NHS London and other global health districts (Texas Medical Center included) in between. Itâs sold out, but the waitlist is open, so itâs worth a punt if you want to know where the money in London health is heading.
Frontiers Health
đïž 20â21 Oct
đ đ©đȘ bcc Berlin Congress Center, Berlin.
Health equity is one of the four main themes, which Iâd like to see more events copy. If youâre already in Berlin this week (and hopefully heading to Berghain after), youâve got good taste (startup passes are âŹ495 + VAT).
AI in Health Summit
đïž 4 Nov
đ đŹđ§ Royal College of Physicians, London.
The Economistâs take on why AI keeps stalling between the pilot and the ward, with trust and accountability on the agenda. Good timing, given the government just pushed liability back to a working group.
Visit the SomX events page or subscribe to SomXâs events roundup for everything else.
Opportunities đ”ïžââïž
đŽ Medical Device Auditor, Scarlet (London). If the governmentâs AI reply left you wondering who is checking these devices, itâs companies like Scarlet, which is authorised to certify them and says itâs shrinking AI update cycles from months to weeks (with the change-control guidance due in December, timely). Youâd be auditing manufacturersâ quality systems against ISO 13485 and EU/UK MDR, so it suits someone with four-plus years in medical devices and some software know-how. And the red pen is presumably included.
đ Head of Clinical Research, Numan (Remote, UK). A 12-month mat cover leading Numanâs real-world evidence on obesity treatment, with oral GLP-1s at the top of the list. If the nail study left you wanting better evidence, hereâs your chance to produce it (PhD and PI experience needed). Itâs fully remote, so your cuticles can stay at home.
𩞠Clinical Data & Device Specialist, Sava (London). Fresh off the ÂŁ27m above, Savaâs whole bet now rides on its pivotal study, and this is the person keeping that studyâs data clean. It suits someone with a few years in medtech clinical ops who knows their EDCs from their eTMFs (and doesnât flinch at the word lock). Weâre 'already tracking, and now weâre getting a filament ten times shorter. Is it that we hate wearing tech, or that we hate the needle? And is it weird that I kind of like injections?
đŁ Want something in front of our 8,000+ readers?
You can break your news as the main story, feature a podcast, job or event... and we even have a few sponsorship packages to choose from...
We review everything editorially - if it is not right for our audience, weâll let you know - everything has to be valuable to this glorious community.
See you next week, Pigeon fans âđŒ

Healthtech Pigeon is produced by SomX - the media-led communications and creative agency for healthcare. We help healthtech, biotech, pharma and public sector organisations tell sharper stories through strategy, PR, content, design, events and media production. Built by clinicians, scientists and creatives who know the space. Work with us.






