Microsoft's diagnostic AI beat generalist doctors on hard cases — but TechCrunch says most of the gain came from the underlying models, not Microsoft's system
October 26, 2026 · 5 items
Microsoft's diagnostic AI beat generalist doctors on hard cases — but TechCrunch says most of the gain came from the underlying models, not Microsoft's system
TechCrunch · October 22, 2026Practice operations
Microsoft AI chief Mustafa Suleyman is publicly chasing what he calls 'medical superintelligence,' and says the company's AI outperforms doctors on hard cases.
The claim rests on a study published in Science in which Microsoft's new AI diagnostic system beat generalist doctors on hard cases.
TechCrunch's own framing is the part worth keeping: 'most of the improvement seems to come from the AI models underneath, not from Microsoft's system itself' — i.e. the product layer may be adding less than the headline implies.
Reported by Rebecca Bellan, Oct. 22, 2026.
Why it matters for an independent practice: When a patient or a referral partner says 'the AI beats doctors now,' the honest physician answer is the attribution question: beat whom, on what case mix, and is the result coming from the vendor's product or from the frontier model anybody can rent? That distinction is exactly what a practice should ask any diagnostic-AI vendor pitching it this quarter — and it's also the content angle for a physician-authored post that strengthens authority instead of chasing the hype.
PHP flaws found with AI triggered CISA, NSA and FBI patch warnings on Oct. 21 — the same week Quest Diagnostics disclosed a breach to the SEC
Medical Economics · October 23, 2026Regenerative medicine
PHP2Shell is the nickname for two critical flaws in PHP — the language behind WordPress, Drupal, Joomla and a large share of the world's websites. Researchers at Wiz, Calif and OpenAI used AI to show the flaws could let an attacker take over a web server without logging in.
The PHP project released fixes Oct. 21, and CISA, the NSA and the FBI urged organizations to update immediately or take vulnerable systems offline until they can.
Separately, Quest Diagnostics told the Securities and Exchange Commission that on Oct. 21 it learned of a breach — a lab vendor most independent practices send patients to.
Richard Payerchin's piece frames both as reaching directly into the independent physician practice: what to check, what to ask, and what to tell patients now.
Why it matters for an independent practice: Two concrete actions this week: get written confirmation from whoever hosts the practice website that PHP was patched after Oct. 21, and have a one-paragraph answer ready for the front desk when patients ask whether their Quest lab results were exposed. Note this touches the de-emphasized 'practice website stack' shape — it's running because the federal alert and the patient-facing Quest angle make it actionable, not theoretical.
476 ACOs still don't know what they earned or owe for 2025 — Arbital Health's COO argues actuarial AI should score the contract midyear instead
Medical Economics · October 22, 2026Practice operations
The 476 accountable care organizations in the Medicare Shared Savings Program in 2025 won't learn until November whether they earned shared savings or owe losses for that year.
CMS delayed 2025 financial reconciliation until after it finalizes the 2027 Medicare Physician Fee Schedule, so it can apply a proposed retroactive guardrail to the projected spending trend it uses to update ACO benchmarks.
CMS also reopened 2024 reconciliation for ACOs whose agreements began that year, replacing results first published in September 2025 with updated figures in July.
Najib Jai, M.D., MBA, chief operating officer of Arbital Health, spoke with Medical Economics about why settlements turn into disputes and how actuarial AI tracks contract performance while the year is still underway.
Why it matters for an independent practice: Any independent group in a risk or shared-savings contract is flying blind for 12–24 months on money it has already earned or lost. The applicable move is asking whether your payer contracts allow midyear performance estimation at all — because a group that can see the number in June can change behavior, and one that learns it in November can only argue about it.
Anthropic's new Claude Duet is aimed at professionals and small businesses — and is explicitly bad at errands
TechCrunch · October 22, 2026Buildable AI
Anthropic says Duet is meant to cater to professionals and small businesses, helping them automate and manage tasks — a different target than consumer assistants like Meta Muse, OpenAI Dots, Hark, Instinct and Grok Bot.
The company draws the line itself: Duet isn't good at handling errands or booking tickets, and it can't show you its browser use as it happens.
What it does do: build a daily briefing, create dashboards, and suggest actions based on the user's data.
It's available through a waitlist for Pro and Max plan users; reported by Ivan Mehta, Oct. 22, 2026.
Why it matters for an independent practice: The 'daily briefing plus dashboard plus suggested actions off your own data' shape is the practice-manager pattern: new-patient inquiries, no-show rate, review volume, ad spend, surfaced every morning without anyone pulling a report. The caveat Anthropic volunteers is the one that matters for adoption — it can't show you its browser use as it happens, so decide what data it's allowed to touch before you connect anything near PHI.
INmune Bio is pointing a quantum machine and AI at its own trial data to find new diseases for an Alzheimer's drug
Longevity.Technology · October 23, 2026Regenerative medicine
Florida biotech INmune Bio is collaborating with Quantum Computing Inc. (QCi), a New Jersey company building quantum machines powered by light; QCi's Dirac-3S will do the heavy lifting.
The approach inverts the usual order — rather than picking a disease and hunting a molecule, the company already holds XPro (XPro1595) and wants to work out what else it could treat.
XPro is designed to block soluble tumor necrosis factor (TNF), quieting damaging inflammatory signaling while leaving other TNF functions the immune system relies on intact; INmune Bio is looking at neurological and rare diseases.
The inputs are INmune Bio's own pre-clinical data plus four clinical trials, combined with publicly available biomedical datasets, analyzed with quantum computing, AI and conventional analytics together.
Why it matters for an independent practice: For regen-med and longevity clinics, this is the credibility frame to watch: 'AI found a new indication' is a hypothesis-generation claim, not an evidence claim, and the gap between the two is where clinics get into trouble with boards and with patients. Useful as a patient-education talking point about how drug repurposing actually works — and as a reminder to never let a vendor's AI-derived indication list become your marketing copy.
