Robin Blythe
@rbly.bsky.social
Assistant professor at Duke-NUS medical school. Mostly interested in health economics, biostats and clinical informatics.
Reposted by Robin Blythe
OHMYGOD
In 2014 I proposed a post Turing Test benchmark! www.smbc-comics.com/comic/2014-0...
No current frontier models Ai come CLOSE to passing this!
In 2014 I proposed a post Turing Test benchmark! www.smbc-comics.com/comic/2014-0...
No current frontier models Ai come CLOSE to passing this!
Saturday Morning Breakfast Cereal - 2014-06-21
Saturday Morning Breakfast Cereal - 2014-06-21
www.smbc-comics.com
February 5, 2026 at 11:48 PM
OHMYGOD
In 2014 I proposed a post Turing Test benchmark! www.smbc-comics.com/comic/2014-0...
No current frontier models Ai come CLOSE to passing this!
In 2014 I proposed a post Turing Test benchmark! www.smbc-comics.com/comic/2014-0...
No current frontier models Ai come CLOSE to passing this!
The last thing you see leaving the clinic after-hours:
February 6, 2026 at 1:42 AM
The last thing you see leaving the clinic after-hours:
I am utilising 100% of my brain by exerting the willpower NOT to blow $500 on a Ninja Slushi
February 4, 2026 at 7:15 AM
I am utilising 100% of my brain by exerting the willpower NOT to blow $500 on a Ninja Slushi
Getting to teach #healtheconomics to motivated health professionals is always so rewarding, genuinely a pleasure to spend time with such an engaged group. Wonderful hospitality, food, and friendliness in Sri Lanka. Even ended up on the #cricket jumbotron, lol
February 4, 2026 at 1:25 AM
Getting to teach #healtheconomics to motivated health professionals is always so rewarding, genuinely a pleasure to spend time with such an engaged group. Wonderful hospitality, food, and friendliness in Sri Lanka. Even ended up on the #cricket jumbotron, lol
Reposted by Robin Blythe
New draft: "Decline effects, statistical artifacts, and a meta-analytic paradox". In this manuscript I show how a common practice in meta-analysis (eg the 2015 Open Science Collaboration) creates artifactual signatures of poor scientific behavior. PDF: raw.githubusercontent.com/richarddmore... 1/x
February 2, 2026 at 2:56 PM
New draft: "Decline effects, statistical artifacts, and a meta-analytic paradox". In this manuscript I show how a common practice in meta-analysis (eg the 2015 Open Science Collaboration) creates artifactual signatures of poor scientific behavior. PDF: raw.githubusercontent.com/richarddmore... 1/x
Congrats to @aidybarnett.bsky.social, Baptiste and team on this important #metaresearch paper. I got to see the early analysis stages of this work and have been waiting with baited breath for it to be published:
www.bmj.com/content/392/...
www.bmj.com/content/392/...
Machine learning based screening of potential paper mill publications in cancer research: methodological and cross sectional study
Objectives To train and validate a machine learning model to distinguish paper mill publications from genuine cancer research articles, and to screen the cancer research literature to assess the preva...
www.bmj.com
January 30, 2026 at 5:25 AM
Congrats to @aidybarnett.bsky.social, Baptiste and team on this important #metaresearch paper. I got to see the early analysis stages of this work and have been waiting with baited breath for it to be published:
www.bmj.com/content/392/...
www.bmj.com/content/392/...
Incidentally, DAGs are a nice tool to figure out when your dataset is completely useless for your research question and you need to just stop. Unfortunately this is when some causal inference studies double down that they've found "truth"
January 30, 2026 at 1:58 AM
Incidentally, DAGs are a nice tool to figure out when your dataset is completely useless for your research question and you need to just stop. Unfortunately this is when some causal inference studies double down that they've found "truth"
Sri Lanka is a stunning place, and the food is great. Plus, there's this amazing dense fog out at sea that reminds me of The Pale from Disco Elysium
January 29, 2026 at 12:08 PM
Sri Lanka is a stunning place, and the food is great. Plus, there's this amazing dense fog out at sea that reminds me of The Pale from Disco Elysium
Gave a fun talk last week on how we should view clinical decision support as behavioural interventions and not medical ones. A key distinction: we need to analyse (and collect!) proximal outcomes like Tx orders rather than always looking at patient outcomes to judge effectiveness #statsky #episky
January 29, 2026 at 11:31 AM
I'm preemptively sorry for this
January 23, 2026 at 11:47 AM
I'm preemptively sorry for this
QTing to not hijack, but the end mechanism seems less important than asking why we even fund research. There has to be a purpose -- even one as vague as advancing knowledge and science as a public good. I just wonder whether the exact means of disbursement should be based on why we even do this.
Can we start seriously discussing a different grant funding system:
Some ideas:
* universal minimum bursary
* lottery
* grants based on the person (not the project)
* judging applicants based on how they support and develop other people
Some ideas:
* universal minimum bursary
* lottery
* grants based on the person (not the project)
* judging applicants based on how they support and develop other people
THE FUTURE OF GRANT WRITING IS HERE!
January 22, 2026 at 2:28 PM
QTing to not hijack, but the end mechanism seems less important than asking why we even fund research. There has to be a purpose -- even one as vague as advancing knowledge and science as a public good. I just wonder whether the exact means of disbursement should be based on why we even do this.
Update: the paper was rejected, lol
Reviewed a paper asking the authors to either provide the data in a public repo or justify why it was only available upon request. They recanted and said actually, the data were owned by a private company, so it was just flat out unavailable 😂😭😂😭
#openscience
#openscience
January 21, 2026 at 4:35 AM
Update: the paper was rejected, lol
I have sympathy for Scimago having to deal with all those messages from confused people trying to submit papers to the wrong journal, but hosting ads like this does a lot to make that sympathy evaporate
#academia
#academia
January 20, 2026 at 2:22 AM
I have sympathy for Scimago having to deal with all those messages from confused people trying to submit papers to the wrong journal, but hosting ads like this does a lot to make that sympathy evaporate
#academia
#academia
Been trawling through the literature on model/AI-assisted decision support and the evidence is crap. Systematic reviews aggregate and launder the crap, giving the impression these things usually work. They usually don't. Some subprime mortgage crisis parallels here. (1/3) #statsky #medsky #episky
January 20, 2026 at 1:58 AM
Reposted by Robin Blythe
Less is more. This aphorism is constantly on my mind. The scientific publication system is under tremendous strain, the last thing it needs is a massive dump of LLM generated/aided papers. If you love science, then show it some love. www.nature.com/articles/d41...
I’m going to halve my publication output. You should consider slow science, too
If we don’t slow down, the research enterprise is going to crash, argues Adrian Barnett.
www.nature.com
January 19, 2026 at 9:13 PM
Less is more. This aphorism is constantly on my mind. The scientific publication system is under tremendous strain, the last thing it needs is a massive dump of LLM generated/aided papers. If you love science, then show it some love. www.nature.com/articles/d41...
Reposted by Robin Blythe
Hey Aotearoa/New Zealand academics: Here's something good you can do today!
Only 1 of 8 universities in AoNZ has left X. I think it should be 8 of 8.
(So far 10 of 37 Australian public unis left X - you're behind 😬)
Full grade report for AoNZ unis, and what you can do to help, follows.
Only 1 of 8 universities in AoNZ has left X. I think it should be 8 of 8.
(So far 10 of 37 Australian public unis left X - you're behind 😬)
Full grade report for AoNZ unis, and what you can do to help, follows.
Want to do something good for change today?
➡️Are you an academic, a student, or an alumna/alumnus of an Australian University?
Of 37 Australian public universities, 9 have already deleted or inactivated their accounts on X. I don’t see why we can’t get that to 100%!
(See end for what you can do)
➡️Are you an academic, a student, or an alumna/alumnus of an Australian University?
Of 37 Australian public universities, 9 have already deleted or inactivated their accounts on X. I don’t see why we can’t get that to 100%!
(See end for what you can do)
January 18, 2026 at 11:01 PM
Hey Aotearoa/New Zealand academics: Here's something good you can do today!
Only 1 of 8 universities in AoNZ has left X. I think it should be 8 of 8.
(So far 10 of 37 Australian public unis left X - you're behind 😬)
Full grade report for AoNZ unis, and what you can do to help, follows.
Only 1 of 8 universities in AoNZ has left X. I think it should be 8 of 8.
(So far 10 of 37 Australian public unis left X - you're behind 😬)
Full grade report for AoNZ unis, and what you can do to help, follows.
One of the best things about the FDA was that it had a fairly global reach in terms of forcing wellness products to put up or shut up if they wanted US market access. A NZ "nootropic" drink was actually removed from shelves for a few months because they claimed proven benefits and the FDA (1/3)
The wellness industry is predicated on poorly defined and often unmeasurable outcomes, though. The benefits I've seen extolled range from improved mood and cognition all the way through to increased libido. Peer-reviewed research is necessary but it's not part of the calculus for these people.
January 16, 2026 at 2:58 AM
One of the best things about the FDA was that it had a fairly global reach in terms of forcing wellness products to put up or shut up if they wanted US market access. A NZ "nootropic" drink was actually removed from shelves for a few months because they claimed proven benefits and the FDA (1/3)
There are many nice things about living in Singapore but one of the best is that this is a $3 breakfast!
January 15, 2026 at 1:30 AM
There are many nice things about living in Singapore but one of the best is that this is a $3 breakfast!
American data professionals, this is your chance to secretly embed Bayesian thinking (hedonistic, fickle, corrupted by epistemic irresponsibility) into American government frequentist institutions (noble, objective, mandated by heaven)
#statsky #rstats
#statsky #rstats
The amusing alternative is that they don't know this is existing practice and someone has convinced them that Bayes is actually AI and is thus aligned with administrative diktat
January 14, 2026 at 3:32 AM
Reviewed a paper asking the authors to either provide the data in a public repo or justify why it was only available upon request. They recanted and said actually, the data were owned by a private company, so it was just flat out unavailable 😂😭😂😭
#openscience
#openscience
January 12, 2026 at 2:39 PM
Reviewed a paper asking the authors to either provide the data in a public repo or justify why it was only available upon request. They recanted and said actually, the data were owned by a private company, so it was just flat out unavailable 😂😭😂😭
#openscience
#openscience
An observation: I find peer review immensely helpful, but not just because I (might) get good comments. Rather, after 6 months waiting, I've learned so much and had so much time to reconsider a paper's methods and application that I get a second crack at improving it.
#academicsky #statsky #econsky
#academicsky #statsky #econsky
January 12, 2026 at 1:28 PM
An observation: I find peer review immensely helpful, but not just because I (might) get good comments. Rather, after 6 months waiting, I've learned so much and had so much time to reconsider a paper's methods and application that I get a second crack at improving it.
#academicsky #statsky #econsky
#academicsky #statsky #econsky
A huge QoL improvement for me, in my boundless ignorance, would be a progress bar for parallel processing when you have passed, say, 10 for-loops to 10 workers. Reporting when each worker is done is fairly useless when you really want to know each worker's progress along the loop
#rstats
#rstats
January 8, 2026 at 10:54 AM
A huge QoL improvement for me, in my boundless ignorance, would be a progress bar for parallel processing when you have passed, say, 10 for-loops to 10 workers. Reporting when each worker is done is fairly useless when you really want to know each worker's progress along the loop
#rstats
#rstats
The irony for me is that most research is a trade-off between all the cool stuff you don't have time to try, and all the boring paid work you need to do in order to meet funding reqs. Like, ideas are all I've got! Thinking of fun things is what gets me out of bed!
So, I just got one of those emails asking me to evaluate an llm's ideas about one of my papers in return for being able to pursue them. Other recipients are offended; I am just baffled. Is 'having ideas' supposed to be some kind of research bottleneck? Maybe some people are in the wrong profession.
January 6, 2026 at 10:55 PM
The irony for me is that most research is a trade-off between all the cool stuff you don't have time to try, and all the boring paid work you need to do in order to meet funding reqs. Like, ideas are all I've got! Thinking of fun things is what gets me out of bed!
Good thread. My concern is, CI is presented as a solution to bad stats when in reality it suffers from many of the same limitations. We can almost never be certain of most reqs in any case -- exogeneity; data integrity; missingness mechanisms; selection bias to name a few (1/2)
#statsky #econsky
#statsky #econsky
Some half-formed thoughts on causal inference bc I've been reading about it very casually. (fucking yes, managed to spell them both right first go)
I get similar feelings to when I was reading about how the replication crisis would be solved by pre-registration and so
I get similar feelings to when I was reading about how the replication crisis would be solved by pre-registration and so
January 4, 2026 at 10:30 PM