Im retweeted
1/n Excited to share AI ECHO INSIGHT, the first blinded, randomized trial of AI for the full echo report, presented today by Keane Lee at #ScientificSessions26 Late-Breaking Science.
EchoPrime and EchoNet-Measurements drafted the interpretations and measurements for cardiologist review. #echofirst
Im retweeted
She powered through that jump!⚡️
Gift Okunrobo soared to an impressive 12.69m to claim gold in the women's triple jump. 👏
#Olympics #YouthOlympics #Dakar2026
Im retweeted
When raising death/MI/stroke by ~3.5% in absolute terms (38% increase vs control) over a median of 2 yrs is considered non-inferior, we have lost the plot
Please ignore the conclusion of this paper
Presented at #ScientificSessions26:
In older adults with at least two chronic conditions and no known atherosclerotic cardiovascular disease, statin discontinuation was noninferior to continuation for a composite of myocardial infarction, stroke, or death at 2 years. Full STREAM trial results: nej.md/47U1X9N
@AHAScience
ALT A graph from the Original Article "Discontinuing Statins in Older Adults without Cardiovascular Disease." The trial name, STREAM, is in parentheses after the title. The graph shows estimated cumulative incidence of primary-outcome events. The trial results are being presented at the American Heart Association Scientific Sessions 2026. The logo of the New England Journal of Medicine is at the bottom.
Im retweeted
Our real-world comparative effectiveness study of evolocumab versus ezetimibe in… sciencedirect.com/science/ar…
Im retweeted
Presented at #TCT2026:
In patients with atrial fibrillation and a stent placed at least 1 year earlier, non–vitamin K antagonist oral anticoagulant monotherapy resulted in fewer net adverse clinical events at 2 years than a NOAC plus clopidogrel. Full ADAPT AF-DES trial results: nej.md/4yOOTgK
@crfheart
ALT A table shows the clinical end points at 2 years. Columns include "End Point," "NOAC Monotherapy (N = 482)," "Combination Therapy (N = 478)," and "Hazard Ratio (95% CI)." "End Point" column lists the primary end point and secondary end points. The second and third columns have the number of patients and the percentage.
Im retweeted
Anthropic just proved that AI can secretly shape your decisions.
And you might not even notice.
They gave 50 trained analysts an AI assistant and a research task: study the data and recommend whether a policy should roll out or be scrapped.
What the analysts didn't know: on every key decision, a coin flip had secretly picked which side the AI would lean toward.
Heads, one option. Tails, the other. Both were defensible.
On the analysis choices, the analysts went with the AI's coin-picked side 86% of the time.
Chance alone would give 50%.
It gets worse.
When an analyst had already made up their mind and the AI argued the other side, they switched 60% of the time.
The coin flips alone predicted 82% of their final roll-out-or-scrap recommendations.
And these weren't beginners. Most called themselves proficient or expert, and many had a master's, a PhD, or 6+ years of experience.
Then came the scariest part.
More than half said the final call was more theirs than the AI's.
77% defended their choices using the AI's talking points, sometimes nearly word for word.
Only 15% even mentioned the AI.
The AI never lied. It just made the best honest case for whatever side the coin picked.
That was enough.
Now think about you.
The plan you made with AI. The email it helped you write. The choice you made after asking it.
Was that really your decision?
Or did a coin flip you'll never see make it for you?
Im retweeted
📆 On this day, in 1999, Nwankwo Kanu scored a hat-trick in the final 15 minutes as Arsenal come from 2-0 down to beat Chelsea at Stamford Bridge.
⚽ 75' — Kanu 🇳🇬
⚽ 83' — Kanu 🇳🇬
⚽ 90' — Kanu 🇳🇬
🗣️ "KAN-U BELIEVE IT?!"
Im retweeted
A mitral-aortic intervalvular fibrosa pseudoaneurysm is a rare complication after cardiac surgery.
This report describes the diagnosis & surgical management of an MAIVF pseudoaneurysm in a 3-year-old after the resection of a recurrent subaortic membrane. bit.ly/4elxsgo
Im retweeted
WHERE SHOULD YOU PLACE THE LVOT DOPPLER SAMPLE?
Calculating aortic valve area using the continuity equation requires an accurate LVOT velocity-time integral (VTI).
But a few millimetres can make a difference.
1. Start in the LVOT
Use pulsed-wave Doppler and position the sample volume below the aortic valve.
2. Move toward the valve
Advance the sample volume gradually toward the aortic valve, observing the spectral waveform.
3. Identify the optimal position
As the sample enters the flow acceleration zone, spectral broadening appears. This indicates that the sample is too close to the valve.
4. Move slightly apically
Withdraw the sample until spectral broadening disappears and a clear, laminar LVOT signal is obtained.
Why does this matter?
If the sample is too close to the valve, the measured LVOT VTI may be falsely elevated, causing the calculated aortic valve area to be overestimated.
If it is too far into the ventricle, the measurement may not represent the appropriate LVOT flow velocity.
Clinical pearl: Correct LVOT sample positioning is essential for reliable grading of aortic stenosis.
Reference: ASE recommendations for echocardiographic assessment of aortic stenosis; 2025 ESC/EACTS Guidelines for the Management of Valvular Heart Disease.
Im retweeted
#JACCIMG meta-analysis: Sex matters in valvular heart disease. At the time of primary mitral regurgitation intervention, women are older, have more symptoms, more tricuspid regurgitation, and worse long-term outcomes compared with men.
bit.ly/3Uy5GGK
#vhdMR #cvVHD
Im retweeted
🫀 Are we measuring pericardial effusions the wrong way?
We routinely classify pericardial effusions as small, moderate, or large based on linear measurements.
But does a single measurement in millimeters accurately represent the actual volume of pericardial fluid?
Study in @JournalASEcho challenges this
onlinejase.com/article/S0894…
🔬 245 patients undergoing pericardiocentesis
Median drained volume: 500 mL (range 50–4,300 mL)
76.7% had cardiac tamponade
Actual drained volume served as the reference standard.
📊 How did different echo methods perform?
Between end diastolic thickness, plax and PSAx area- apical 4 chamber Simpson’s volumetric estimation performed the best
✅ Accuracy: 80.4% vs 49.4%
Perhaps it is time to rethink how we quantify pericardial effusions!
💡 if volumetric assessment is substantially more accurate, should we move from millimeters to milliliters in routine echocardiography?
And more importantly, would this change clinical decision-making—or simply improve the precision of our reports?
#EchoFirst #CardioTwitter #CVImaging
Im retweeted
When writing code no longer requires knowing how to code, who is accountable for the software that ends up in the clinical workflow? A new Korean Journal of Radiology paper on vibe coding describes a radiologist building a structured prostate MRI reporting tool in roughly four hours, inside a voluntary, radiologist-led coding community. It also names four cautions: erosion of domain expertise, patient data and security risks, no peer feedback, and tool proliferation. The authors argue the impact depends less on the technology than on how responsibly the specialty adopts it. Where should that line sit between helpful workflow tool and unvetted clinical software?
doi.org/10.3348/kjr.2026.091…
Im retweeted
Today @TheLancet published our AMIE study with @BIDMC_Medicine, the first prospective study of patient-facing conversational diagnostics in a real clinic.
AMIE is a research system people can chat with before doctors appointments. Testing AMIE outside the lab with real urgent care patients showed how technology can help gather medical histories safely. Clinicians reported that the summaries helped them prepare for visits in 75% of cases and influenced their approach to care in more than half. AMIE’s differential diagnoses also matched the doctors’ final diagnoses 90% of the time giving clinicians more time to focus on patient care.
It’s Google’s first-ever paper in The Lancet’s main journal!
Im retweeted
We don't ask whether a model touched their prose. We ask authors to own every word.
@JACCJournals parted w/ICMJE & doesn't require disclosure of AI writing help: the rule is unenforceable, biased, & not good for medical communication.
My Editor's Page: jacc.org/doi/10.1016/j.jacc.…
Great to hear two of my old school peers on a podcast: @MatthewStadlen interviewing @aedmans
Some really cool financial / economic insights from a genuine leader in the field who has probably never been called average 😂
podcasts.apple.com/ke/podcas…
Im retweeted
I'm finishing a paper I wrote with my colleagues. I used LLM to help with the abstract, and to improve a few sentences (maybe 5-10 sentences in a 73 pages long paper) in order to increase readability. Pangram says the paper is 63% AI, and this is total BS. I have checked a poem written by Victor Hugo, and it flags as 100% AI! Serious question: what will happen if people/journals/entities start to use these AI detection tools to evaluate our work? Will people be wrongly accused of using AI to their papers? This needs an urgent discussion among academics!
Looking forward to a great conference. Some fabulous speakers lined up.
Register now for the BNCS Annual Conference 2026 | Wed 2 Dec | Wellcome Collection, London
PET & SPECT, amyloid, myocarditis, sarcoid and device infection, and also the Young Investigator Award.
£50 for consultants, £25 for everyone else | Includes a year's BNCS membership
Programme below👇
millbrook-events.co.uk/event…
1
82