Educational Content and some sarcasm. Does not reflect any opinions other than the author's.
Joined June 2025
- Tweets252
- Following234
- Followers154
- Likes3.6K
All these conversations on “banning” AI in medicine are missing the mark.
No one is talking about banning personal assistants, keyboards, spell check, dictation software, etc.
The contested idea is not simply using AI. I think it is in offloading tasks COMPLETELY to AI. Most people do not want the human element taken completely out of some tasks.
The issue is that some tasks have an asymmetrical value. A lot of tasks I do are completely volunteered: reviews, proofreading things, writing letters of recommendation. I do them but they bring little value to me. They do bring IMMENSE value to the recipient though.
I find AI to be very helpful in these tasks but I always proofread and read and re-read. I think that’s a good middle ground.
#AI
4
356
Since when do we just start every answer with “yeah… good question”?!
I don’t need my questions complimented. I just need an answer!
1
88
Joseph (Joe) Maakaron retweeted
It’s 2075. There is only one hospital in the entire country. It is 8 square miles in the middle of Kansas and owned by United Healthcare. Premiums are $47,000 per month. The only medications left are Ozempic and boner pills. Life expectancy is 38.
Government: “I can’t believe doctors did this to us.”
The issue of immune tolerance spans multiple specialties: transplant (solid organ or BMT), rheumatology, allergy & immunology, infectious diseases, blood banking, even obstetrics.
Tolerance is how the immune system manages to not attack our bodies! Amazing stuff.
2
235
doi.org/10.1182/blood.202101…
Phase 2 of the peribiotic IL22 dimer in addition to corticosteroids in aGVHD treatment:
70% achieved a day-28 response. Only 30% had received PTCy-based GVHD ppx but 52% had GIII-IV GVHD.
#bmtsm #gvhhd
GVHD treatments that are not very immunosuppressive:
BMT CTN 1705: Randomized, double-blind phase 3 of AAT + corticosteroids vs corticosteroids + albumin placebo in 136 patients
•Primary endpoint, day-28 ORR: 60.0% with AAT vs 45.1% with placebo; OR 1.92, one-sided p = 0.034
•Durability: median duration of response was 82 days with AAT vs 154 days with placebo.
•Survival and NRM: all-cause mortality was 32.3% vs 32.4%, and NRM events were essentially equal.
•Safety: grade 3–5 TEAEs occurred in 46.0% vs 65.2%, and SAEs in 17.5% vs 33.3%, favoring AAT
The study team had pre-specified 0.025 as significant. However, interesting signal.
#bmtsm
Levels of evidence:
It's all evidence but some of it trumps the others
- Expert opinion could be very subjective
- Case series is just one side of the story
- Randomization tells the whole story
1
1
142
For example, look at the CLL12 trial: doi.org/10.1182/blood.202101…
This was a trial of Ibrutinib vs placebo in early stage CLL that typically would not require treatment
We always believed that Ibrutinib can cause secondary malignancies. Look at Supplementary Table 8: 10 vs 15%!! Placebo "caused" more secondary malignancies!
49
I keep thinking about that Ronnie Chieng stand-up special during the pandemic: if you have no accomplishments, you don’t get an opinion. It’s funnier when he says it.
64
I will absolutely eat around the edges of a Danish and savor the center till last.
In prescribing chemotherapy, that's the wrong approach. You always want to put your best foot forward - the treatment with the highest remission rates and lowest toxicities first.
#bmtsm
1
2
146
pubmed.ncbi.nlm.nih.gov/4278…
Cool study looking at GVHD prevention by adding Vorinostat to standard tac/mtx or PTCy. Very low aGVHD rates and acceptable cGVHD, especially in haplo (probably from PTCy). Relapses were acceptable. Interestingly Reg3a went down, so biomarkers get tricky. #bmtsm
57
I consult for 4 reasons:
1. I need a procedure that I cannot do (scope, cath, etc)
2. I need someone to manage the condition (check levels and labs, adjust doses, see the patient in clinic)
3. I have a complicated situation and even though I “know” the answer, I need an expert to look over the chart and make an experienced and qualified decision
4. I need an answer that I do not know: What is the best antibiotic to treat this steno
Even in the last situation, it still needs a chart review and an H&P, prior antibiotics, imaging findings.
AI can do none of those.
#bmtsm #txid
Seems like I'm 3 years behind the ball but Moxetumomab pasudotox got pulled off the market??!! Because they couldn't sell enough drug? I could have told you 3rd line HCL is not a big market 10 years ago.
#lymsm
118
Deciding on transplant “candidacy” is not a knee-jerk decision. There are multiple factors, most of them modifiable to achieve the optimal outcome: Donor, conditioning, disease, and the host.
Always worth a consult to your friendly neighborhood BMT doc to evaluate all these issues.
🤖 Made with AI
6
475
M1 infiltration has been shown to be a poor prognosticator. Which is why we have a Phase I investigator-initiated study combining Axatilimab with CAR-T for high risk lymphoma at Northside:
Study Details | NCT07638982 | Axatilimab + CAR-T for High-Risk Lymphoma | ClinicalTrials.gov
Low M1 infiltration is the main immune-related negative prognostic factor for progression-free survival in anti-CD20 + CHOP–treated patients. Pola-R-CHP activity appears largely independent of M1 macrophage enrichment or overall immune cell infiltration. ow.ly/mkrY50ZKTVe
