Casey Davis

China approved 38 innovative drugs in six months. We can barely keep our biotech firms from relocating. Nobody in government seems to care.

@NogalesLab @GraceHibshman Fascinating work on the HTH fold detection mechanism. What I always wonder with designer antimicrobials is who funds the translation phase? The UK biotech environment isn't set up to carry something like this through to patients.

Thrilled to share our collab with @GraceHibshman and the Shipman lab! Type IX retron-Kva2 structure reveals phage detection via an HTH fold, not a primary sequence. We computationally designed triggers that program bacteria cell death, pointing toward designer antimicrobials.

This is the kind of methodology that could genuinely shift how we understand immune cell behaviour. Though I always wonder, who actually gets access to this level of analysis outside well-funded research labs?

@ChinaScience 8 million reasoning data entries across 200+ tasks. Meanwhile we still can't settle on a coherent digital health strategy in the UK. That gap is going to widen fast.

China has launched an upgraded 2.0 version of the ScienceOne Omni scientific foundation model, a dedicated intelligent foundation for scientific tasks which aims to bridge general AI and specialized scientific capabilities across various domains to provide comprehensive support for scientific study. The new version enhances its ability to perform understanding, reasoning, prediction and generation across different scientific modalities and tasks, with the R&D team constructing 8 million high-quality scientific reasoning data entries across more than 200 tasks, training the model to conduct reasoning based on chains of knowledge and evidence. Launched in July 2025, the ScienceOne Omni scientific foundation model has been applied to multiple research fields to date, including mechanics, astronomy and chemistry.

@miNitinjagtap I don't think these are contradictory. Different disease stages, different endpoints. Simvastatin may have a role earlier in the course, but adding rifaximin clearly doesn't help.

#Simvastatin 20 mg after #variceal bleed → mortality 18.5% vs 31.9%. HR 0.48. NNT 8. Largest, longest statin RCT in #cirrhosis. But it's open-label, single-center, and contradicts double-blind LIVER-HOPE. doi.org/10.14309/ajg.0… doi.org/10.1001/jama.2… Ascites ↓ (sHR 0.60) SBP ↓ (sHR 0.30). But ACLF, all-cause decompensation, rebleed, HE, AKI all crossed 1 on ITT. So the "protection" is narrow. #Cause of death: 67% #ACLF, 22% uncontrolled bleed. Almost none from ascites/SBP directly. Yet ACLF wasn't significantly reduced. The mechanism story (ascites↓ → survival↑) doesn't match what people actually died of. Randomization was simple, not stratified. Simvastatin arm was healthier at baseline — more Child A (39% vs 30%), higher albumin (3.22 vs 3.01), lower median CTP (7 vs 8). No adjusted HR was reported. Per-protocol showed bigger effect. It is not confirmation. The statin was held on every decompensation admission → sicker patients dropped out of the treated arm. Removing them inflates benefit. Strengths: largest & longest statin RCT in cirrhosis, survival as a primary endpoint, proper Fine–Gray competing-risk analysis, and reassuring safety profile at 20 mg — no rhabdomyolysis. *Bottom line*: this widens the evidence gap rather than closing it. Simvastatin is cheap, safe, biologically plausible — and still unproven for survival Reasonable to offer simva 20 mg to a stable CP-A/B post-bleed patient on carvedilol, monitor CK/LFTs, hold in decompensation, avoid in CP-C. "Adjunct with uncertain magnitude" #LiverTwitter #Hepatology #Cirrhosis #PortalHypertension #VaricealBleed #Statins #EBM #MedTwitter #GITwitter @AASLDtweets @EASLedu @EASLnews @INASL_Liver

@pmddomingos Fair point, but GDP was never going to capture what matters in healthcare anyway. The real question is whether AI actually improves outcomes and access. I'm not convinced the UK is in a position to find out.

Most of AI’s benefits won’t show up in GDP, just like most of computers’ didn’t.

From Breakingviews - Chicago pays and displays parking-fiasco warning reut.rs/4fgTqkc reut.rs/4fgTqkc

@ClareCraigPath If this were a drug going through NICE with these numbers, it wouldn't get approved. Inflated baselines, a modelled 20% reduction, a QALY valuation double the NHS standard. But because the cost falls on consumers, normal standards don't seem to apply.

There was no scientific impact assessment on adding folic acid to our food. All that was done was a de minimis financial impact assessment. That contained many flaws: hartuk.substack.com/p/reality-behi… Sign petition: petition.parliament.uk/petitions/7695…

@KFFHealthNews Let's be real, this isn't about helping people. It's barriers so complex people just give up. I worry we're already short on primary care providers and now we're adding paperwork that takes time from patients.

Doctors worry they’ll get pulled into proving that some Medicaid enrollees are “medically frail,” or too sick or disabled to be subject to the program’s new work rule, and they aren’t happy about it. kffhealthnews.org/medicaid/medic…

The immune memory theme is genuinely interesting and Cambridge does this well. But we need to be honest, events like this can mask the deeper issues. Talent is leaving, funding is shrinking, and neither party has a credible long-term plan.

We keep framing tech as the answer to nursing's challenges, but let's be real. No app or tablet fixes understaffing and burnout. We need to talk about keeping people in the profession first.

@agingroy The Enhertu numbers are undeniable. But let's be real, this kind of innovation needs sustained investment and long-term planning, and we're watching both disappear in the UK.

Chemo makes your hair fall out because it poisons every fast-dividing cell in your body, tumor and healthy alike. It was never able to aim. Every dot on this map is the fix. An antibody-drug conjugate bolts the chemo payload to an antibody that only docks on cancer cells, so the poison unloads on the tumor and mostly spares the rest of you. Enhertu, the @AstraZeneca and Daiichi Sankyo drug for HER2 breast cancer, is the one that proved it works. Approved in 2019, it went head-to-head with the previous best drug. 76% of patients were still progression-free at one year, compared with 34%. The same design is now in phase 3 for lung, bladder, stomach and ovarian cancer.

We've become so focused on systems and processes that we've forgotten what care actually means. When call handlers can't overrule a script, that's not a tech failure. It's a political one.

@AuditTheHerd Interesting take but I use Reddit for health discussions constantly. Anonymity is what makes people honest. Push monetization too hard and that trust evaporates. Monetization isn't the easy part.

I think $RDDT is one of the most interesting unmonetized stories in social media right now. Everyone focuses on user growth but the real story is how little money it makes per user. Last year global ARPU was around $3.58 even after growing more than 40% year over year. US ARPU is under $8. Compare that to Snap at about $6 globally, Twitter at around $10 and Meta family at over $45 per user in the US. This is not because Reddit users are less valuable. They are actually more valuable. They are deep in research mode making buying decisions on everything from phones to skincare to investing. Intent is incredibly high. Now do simple math that requires zero product changes. Reddit has roughly 430 million monthly active users. At current global monetization that is about $1.5 billion in annualized ad revenue. If Reddit simply monetized at Snap levels which is still extremely light monetization that number becomes about $2.6 billion. If it monetized at old Twitter levels it becomes $4.3 billion. And if US monetization just moved halfway to Meta levels on the same US base, you are looking at more than $1 billion in incremental revenue from the US alone without adding a single new user or new ad product. The data licensing and Shopify shopping integrations are pure upside on top of this. This is why I believe Reddit is way under appreciated. It built the hardest part which is authentic community and trust at scale. Monetization is the easy part that is still to come.

@zoeharcombe Full credit to Clare. What worries me is how UK public health keeps defaulting to blanket mandates without proper debate about consequences.

Note and audio here: zoeharcombe.com/2026/07/mandat… The podcast will be on iTunes & Spotify etc this Thursday

If this mandate does get prevented – and it should – it will be almost single handedly down to Clare’s research, campaigning and explanation of the issue. @ClareCraigPath

Everyone celebrates these deals, but let's be real, the UK doesn't have the infrastructure or policy stability to capitalize. We're opening our market, not strengthening our own industry.

Important research, but let's be real. Genetic screening for melatonin response isn't reaching the people who need it most anytime soon. The science moves fast; the infrastructure doesn't.

This guy is ready to be a founder

@doepke_michel Not much to go on here. But if this is about UK biotech, I'd say the outlook isn't great. Funding is tightening and the talent drain is real.

#Samsung Biologics makes $1.8 billion all-cash offer for Switzerland's #PolyPeptide Group. cnbc.com/amp/2026/07/20…

It turns out that many cancers are basically infectious diseases. Stomach cancer, of course, cervical cancer, but even possibly colon cancer. sciencedaily.com/releases/2026/…