Allan Joseph
@allanmjoseph.bsky.social
I take care of critically ill kids, study the systems in which we take care of them, and try to make those systems better.
TurboTax is about to be integrated into ChatGPT, where customers will be guided through tasks tied to their tax filings or financial profile by the AI chatbot.
TurboTax owner Intuit is also paying OpenAI more than $100 million a year to power AI agents.
TurboTax owner Intuit is also paying OpenAI more than $100 million a year to power AI agents.
November 18, 2025 at 5:37 PM
"I didn't write my documentation" is probably not an airtight defense, either.
November 17, 2025 at 2:45 PM
"I didn't write my documentation" is probably not an airtight defense, either.
My favorite line (as a Notre Dame alum) is when Martin says "ah I hate when Fr. Mike covers Bulldog's show, it's all just Notre Dame, Notre Dame, Notre Dame" because it's also incredibly accurate
November 14, 2025 at 1:34 AM
My favorite line (as a Notre Dame alum) is when Martin says "ah I hate when Fr. Mike covers Bulldog's show, it's all just Notre Dame, Notre Dame, Notre Dame" because it's also incredibly accurate
God I love this show. S4E1 is my favorite episode but it is stiff competition
November 14, 2025 at 1:25 AM
God I love this show. S4E1 is my favorite episode but it is stiff competition
Don't forget the third measure otherwise you might have an existential crisis over your age
November 14, 2025 at 1:25 AM
Don't forget the third measure otherwise you might have an existential crisis over your age
This would be incredibly complicated for me.
November 13, 2025 at 5:11 PM
This would be incredibly complicated for me.
What we need is a critical care trial that uses AI/ML when it could have used logistic regression, with a noninferiority design with a wide margin permitting an NNH of 20 for one additional mortality. Then there will be plenty of takes from her account alone.
November 6, 2025 at 2:34 PM
What we need is a critical care trial that uses AI/ML when it could have used logistic regression, with a noninferiority design with a wide margin permitting an NNH of 20 for one additional mortality. Then there will be plenty of takes from her account alone.
Please consider serving as a job reviewer for HP OfficeJet 3550. You can sign up using EditorialManager, because even office printers have chosen to use this software for some reason.
November 3, 2025 at 7:39 PM
Please consider serving as a job reviewer for HP OfficeJet 3550. You can sign up using EditorialManager, because even office printers have chosen to use this software for some reason.
Tremendously insightful. (Coincidentally, another peds-adult difference I'd love to explore...a HUGE portion of inpatient/ICU resources are spent on a small fraction of kids with medical complexity/tech dependence, but they are also unlikely to die. So there may be more persistence.)
October 29, 2025 at 4:19 PM
Tremendously insightful. (Coincidentally, another peds-adult difference I'd love to explore...a HUGE portion of inpatient/ICU resources are spent on a small fraction of kids with medical complexity/tech dependence, but they are also unlikely to die. So there may be more persistence.)
As usual I recommend @emilymoin.com's ongoing and insightful thread on this RCT
October 29, 2025 at 3:38 PM
As usual I recommend @emilymoin.com's ongoing and insightful thread on this RCT
I also bet that SICUs/transplant ICUs/trauma ICUs have better outcomes and the PICU is basically one giant mixed unit (though in large hospitals the congenital cardiac ICU is separate)
October 29, 2025 at 2:25 PM
I also bet that SICUs/transplant ICUs/trauma ICUs have better outcomes and the PICU is basically one giant mixed unit (though in large hospitals the congenital cardiac ICU is separate)
Ah my (old, heuristic-y) number was 30-50% mortality but those are probably more like tertiary/quaternary MICUs. (I think the general PICU population is like 2% on average but some of that is denominator inflation by admitting kids to PICUs whose equivalents never would have been in MICUs)
October 29, 2025 at 2:23 PM
Ah my (old, heuristic-y) number was 30-50% mortality but those are probably more like tertiary/quaternary MICUs. (I think the general PICU population is like 2% on average but some of that is denominator inflation by admitting kids to PICUs whose equivalents never would have been in MICUs)
100% agree conceptually. The snarky (but correct) answer is "it depends on which rivets and why they crash!" Which is also true for our heterogeneous ICU patients. I also wonder about second-order effects. If we put fewer A-lines in...does our capacity to insert/maintain them worsen over time?
October 29, 2025 at 2:09 PM
100% agree conceptually. The snarky (but correct) answer is "it depends on which rivets and why they crash!" Which is also true for our heterogeneous ICU patients. I also wonder about second-order effects. If we put fewer A-lines in...does our capacity to insert/maintain them worsen over time?
If you want to compare/contrast with the Peds ICU, let's chat. On the one hand, our mortality rate is an order of magnitude lower (so less has to go "right" to survive since 95% of patients do) but on the other my gut instinct is that individual interventions may have higher marginal benefit.
October 29, 2025 at 1:51 PM
If you want to compare/contrast with the Peds ICU, let's chat. On the one hand, our mortality rate is an order of magnitude lower (so less has to go "right" to survive since 95% of patients do) but on the other my gut instinct is that individual interventions may have higher marginal benefit.
Oh the coffee at Constellation rules.
October 25, 2025 at 12:44 PM
Oh the coffee at Constellation rules.
The @criticalcarereviews.com newsletter is a game-changer.
October 14, 2025 at 12:17 AM
The @criticalcarereviews.com newsletter is a game-changer.
Maybe analogous: in '16 the Neonatal Resuscitation guidelines stopped routine intubation/suction for meconium. Delivery room intubations dropped a ton, and what was once a resident job became a fellow job...and I'd bet effective BVM training got worse too.
publications.aap.org/hospitalpedi...
publications.aap.org/hospitalpedi...
Impact of the Revised NRP Meconium Aspiration Guidelines on Term Infant Outcomes
OBJECTIVES:. To evaluate the association of the Neonatal Resuscitation Program, Seventh Edition changes on term infants born with meconium-stained amniotic fluid (MSAF).STUDY DESIGN:. We evaluated the...
publications.aap.org
September 24, 2025 at 7:49 PM
Maybe analogous: in '16 the Neonatal Resuscitation guidelines stopped routine intubation/suction for meconium. Delivery room intubations dropped a ton, and what was once a resident job became a fellow job...and I'd bet effective BVM training got worse too.
publications.aap.org/hospitalpedi...
publications.aap.org/hospitalpedi...
One was a halfback pass, right? Because I think the stat would get more play if it was 5 straight INTs by the QB
September 14, 2025 at 3:08 AM
One was a halfback pass, right? Because I think the stat would get more play if it was 5 straight INTs by the QB
Let me know if you want to think about peds inpatient/specialty care in this project
September 5, 2025 at 6:54 PM
Let me know if you want to think about peds inpatient/specialty care in this project
Very different population but this is something the neo and peds groups have tried to assess, and I'm sure someone will do after Oxy-PICU too. Two examples:
www.nejm.org/doi/full/10....
jamanetwork.com/journals/jam...
www.nejm.org/doi/full/10....
jamanetwork.com/journals/jam...
June 12, 2025 at 9:12 PM
Very different population but this is something the neo and peds groups have tried to assess, and I'm sure someone will do after Oxy-PICU too. Two examples:
www.nejm.org/doi/full/10....
jamanetwork.com/journals/jam...
www.nejm.org/doi/full/10....
jamanetwork.com/journals/jam...
Our institution uses it as a very late pressor: journals.lww.com/ccejournal/f...
We try to stratify by direct renin levels if possible too
We try to stratify by direct renin levels if possible too
June 3, 2025 at 3:01 PM
Our institution uses it as a very late pressor: journals.lww.com/ccejournal/f...
We try to stratify by direct renin levels if possible too
We try to stratify by direct renin levels if possible too