←
Vincent Rajkumar54,435 posts
Vincent Rajkumar avatar✓

Vincent Rajkumar ✓

@VincentRK

Editor-in-Chief, Blood Cancer Journal; Chairman @IMFmyeloma Board; Cancer & Myeloma Research; Opinions solely personal views bit.ly/3f4TnJl

◷ Joined March 2009▧ 4,535 Photos and videos
Vincent Rajkumar avatar
Vincent Rajkumar@VincentRK·10 Feb 2021

I would like to share the story of how a patient with cancer came up with the idea for a randomized trial, & how listening to him saved a lot of lives. 1/ In 2002, I had just completed a randomized trial with the notorious drug thalidomide for the cancer, multiple myeloma.

UpToDate avatar
⟳ Vincent Rajkumar retweeted
UpToDate@UpToDate·Nov 16

UpToDate Expert AI is now a plugin within ChatGPT for Healthcare from OpenAI. Same trusted, expert-authored clinical content. One more place clinicians can reach it. AI will keep changing how clinicians find answers. What the answers are built on still matters most. Check out the announcement below! #UpToDate #HealthcareAI #HLTH2026

Vincent Rajkumar avatar
⟳ Vincent Rajkumar retweeted
Vincent Rajkumar@VincentRK·Nov 9

I’m at the @UpToDate Editors meeting today. I’ve been an author/section editor for 20 years. The breaking news is UpToDate Expert AI. Check out UpToDate Expert AI at your institutions. You should have access if you have access to UpToDate. UpToDate Expert AI is unique. It is AI plus the world’s top experts analysis and interpretation rolled into one. It offers AI’s magic and proficiency with the human expert element added. In my experience AI is incredibly helpful but very few of us rely solely on AI to make clinical decisions for a serious illness. Patients still want input from the human expert. Thats what makes UpToDate Expert AI unique. It has both. That helps the treating physician make the best management choice for their patient.

Post attachment
Evgeniy Padaliak avatar
⟳ Vincent Rajkumar retweeted
Evgeniy Padaliak@genepadaliak·Nov 9

It’s pretty nice being on the AI side of this. @UpToDate's editors and contributors have spent decades reviewing and synthesizing the evidence so we don't need to ask an LLM to decide if a questionable paper should shape the answer. So most of our time goes into tuning how it translates that knowledge for a specific patient: - Cautious, but not so cautious that it delays a decision. - Confident, but without overstating things. - Helpful, but without making broad extrapolations. Finding the balance is hard and I believe the recely launched V2 (US and Canada only) is very good at that. If you’ve tried it, would love to hear what you think.

Post attachment
Shane Welter avatar
⟳ Vincent Rajkumar retweeted
Shane Welter@ShaneWelter·Nov 9

The point about serious illness is the right line to draw. Keeping a named expert's judgment next to the AI answer gives the treating physician something to weigh, not just something to accept, and twenty years of editing is exactly that kind of judgment.

Open Medicine Insights avatar
⟳ Vincent Rajkumar retweeted
Open Medicine Insights@OpenMedInsights·Nov 9

· Open Medicine Algorithm Spotlight · Sharing a practical @OpenMedicineHQ algorithm by @VincentRK “Newly Diagnosed Myeloma” (Last updated November 7, 2026) 👉🏻FIRST: Watch Dr. Rajkumar's intro video on Treatment of Newly Diagnosed Multiple Myeloma youtu.be/Qx16l86GwIg?is=j994… 👉🏻LIVING algorithm link: myopenmedicine.com/algorithm… #OpendMedicine #Myeloma #mmsm @OpenMedKate @OpenMedicineHQ @scserendipity1

Post attachment
Vincent Rajkumar avatar
Vincent Rajkumar@VincentRK·Nov 9

I heard today that UpToDate Expert AI is integrated into Abridge and will be available at point of care providing management guidance for the specific patient being seen. Amazing

Vincent Rajkumar avatar
⟳ Vincent Rajkumar retweeted
Vincent Rajkumar@VincentRK·Nov 7

1) This is why progress is slow. Even if we can sequence the whole genome we are far far away from understanding how it all works. We cannot fix the human being like you can fix a car. In a car you know how every little part works. 2) This is why we need clinical trials and randomized trials to figure out if what we think is the solution to disease actually is the solution and not a confounder. Whether it actually works like we think it may. 3. This is why big randomized trials fail to show benefit even if we think we know what the problem is and we have evidence that the medicine does what it’s supposed to. The latest example: Lp(a) HORIZON Trial - This 7-year, Phase 3 trial of over 8,350 patients where the drug successfully lowered genetic lipoprotein(a) [Lp(a)] levels by up to 80%, but it failed to significantly reduce heart attacks, strokes, or cardiovascular deaths compared to placebo. 4) The need for randomized trials means progress is slow because we need to do the trial and prove that the intervention works, unlike the car mechanic who knows even for the most complicated car how to fix a problem. Randomized trials take years to complete. 5) This is why new drugs are expensive. Because we simple cannot go from bench to clinic. The expensive essential clinical trial is in the middle. 5) This is why while I’m all in on AI and it’s mind bogglingly good, while it can solve centuries old math and physics problems, I think it’s far more difficult to have that same effect in medicine. What we feed the AI as truth is not the truth. What we know in medicine is often a piece of a puzzle which itself is a piece of a bigger puzzle and that’s a piece of a bigger puzzle with all the pieces jumbled together in one massive pile. AI will definitely accelerate the progress wildly. But it’s still hard because the ground truth we tell it is not the real ground truth. 6) This is why we need to be humble. The human body is complex. Biology of life is complex. It’s not like fixing a broken radio although we often approach it as such. See this beautiful essay that’s been published many times. cell.com/fulltext/S1535-6108…

Ajai Chari avatar
⟳ Vincent Rajkumar retweeted
Ajai Chari@AjaiChari·Nov 7

Given risk of early infxns & need for site specific insurance authorization wit possibility of step ups being done at a different site than Ivig ….recommend ivig dose even prior to step up if baseline hypogamma and if not as soon as crs risk resolved regardless of igg level

Samer Al Hadidi, MD,MS,FACP avatar
⟳ Vincent Rajkumar retweeted
Samer Al Hadidi, MD,MS,FACP@HadidiSamer·Nov 7

IVIG Q 28 days to start with BsAb regardless of IgG level and to continue for at least 6 months after discontinuation for those whom we stop therapy in