Chris McMaster
@DrCMcMaster
Followers
1,405
Following
1,680
Media
468
Statuses
4,847
Husband of @natalieytan , doctor of rheumatic diseases, multiplier of matrices.
Melbourne, Australia
Joined February 2018
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Elon
• 1438996 Tweets
Musk
• 1010695 Tweets
無料クーポン
• 271355 Tweets
増量作戦
• 198188 Tweets
#ファミマ増量クランチ
• 57920 Tweets
智弁和歌山
• 43563 Tweets
智弁和歌山
• 43563 Tweets
アジアツアー
• 39818 Tweets
台風8号
• 25946 Tweets
LagunyaZAGAT PasBANGETTT
• 18701 Tweets
GALAUmu GALAUku
• 17969 Tweets
Süleyman Seba
• 17096 Tweets
サブスク解禁
• 16753 Tweets
パート1
• 15716 Tweets
राम रहीम
• 14359 Tweets
KDRT
• 14085 Tweets
Academic papers should all be submitted to a single preprint server and then journals should have to compete on who gets to publish. Submitting to one journal at a time is a waste of everyone's time and energy.
133
1K
12K
Vitamin D is a cult and the medical community has to accept that we were complicit in that cult for 15 years.
160
168
2K
@bryancsk
The myth of Australia’s uninhabitability is just a lie we’re told to mask the endless bureaucracy and utter laziness that props up our oligarchical ruling class. “Australia is a lucky country run by second rate people who share its luck”
18
14
612
1 million Australians learn what "barre" means and someone at Google analytics scratches their head.
10
51
261
@wagieeacc
@SashaMTL
You actually tried to make a quick buck at the expense of patients and it seems like you’re trying it on again.
1
3
187
"Sorry, but if you COVID panic merchants could shut up for a moment, I've got some dumb shit to insinuate here. Here's hoping Rand Paul retweets this one!"
19
10
143
This release has flown under the radar. Very excited about HuggingChat on my phone!
@huggingface
8
29
144
The problem with ChatGPT is that it has emboldened Edward from Omaha, who used it to generate his grocery list, to publicly challenge
@ylecun
about AGI.
11
7
142
@DGlaucomflecken
@AtomsksSanakan
@VPrasadMDMPH
@K_Sheldrick
Just found footage of Vinay responding to this tweet.
0
1
138
I turned off replies very quickly. Lots of anecdotes, some cultists providing zero evidence for their claims. I truly appreciate those who provide anecdotes about feeling better after taking vitamin D – I'm glad things improved. But that is not causal evidence.
5
1
128
Releasing a new OS model for clinical rheumatology - IL-7B. It's a merge of two seperate finetunes using the same recipe/datasets. One model trained from Mistral 7B base and the other from the recent BioMistral weights.
9
39
117
And when I say "turned off" I mean "muted". Words are difficult. Maybe vitamin D supplements will help?
1
1
110
@KetanJ0
"I'm doing my part"
Woah, billionaire can afford responsibility-free parenthood by hiring an army of nannies. Now this is newsworthy 🙄
1
0
108
Devin is amazing. It solves the greatest problem we have in 2024: not enough mediocre apps. More mediocre apps!
4
2
87
@giffmana
Weird paper. They spend much more time talking about what they wouldn’t do, rather than what they actually did.
1
3
75
@bhargreaves
@bobbyfijan
Lame? Is this the language people use? If so, maybe the problem was giving so much money to 12-year-olds.
0
4
69
One little HTML file to effortlessly share ML models that can be run locally without installing python or any packages. No IT issues, no data transmitted over the internet.
@pyscript_dev
is a game changer for collaborating on and validating clinical ML models across institutions.
0
13
66
I know we're meant to hate dimensionality reduction now, but the
@weights_biases
embedding projector is actually a life changing tool
2
8
60
Euclidean distance = liberal
Manhattan distance = conservative
5
8
57
@MaartenvSmeden
Everyone knows you just need to order your X and y independently. Then you're guaranteed the best results 👌
4
0
52
My standard test with these is to ask it about a disease that doesn't exist and watch it confabulate. Good one, Dr Gupta.
3
11
51
@clairlemon
@misha_saul
His original tweet was him bragging about wasting the time of 3 radiologists with a misuse of AI. This tweet is pure cope.
1
0
46
@wagieeacc
@SashaMTL
Call me a cynic, but I question your motivations. Track record and all.
And yes, your website is a UI, SEO and a prompt slapped on top of the OAI API, but those things add the illusion of trust and accountability. It’s not the same as llama.cpp in a terminal.
2
1
47
@VCBrags
Nobody can convince me this strategy would result in less benefit to society than 90% of startups.
1
0
42
Excited to publish our review of deep learning for rheumatologists in A&R (early and open access now – pretty typesetting pending)
@ACR_Journals
3
10
42
@RobertdeGraaf2
@bryancsk
This is a straw man. There’s 25,000km of sparsely populated coastline before we need to start reaching for the Simpson desert.
3
0
38
@tpq__
I use a different virtual environment for every project. By the end, every virtual environment contains the same 30+ packages and I'm left with an empty feeling deep within my soul.
0
3
39
@KanakaMPH
The correct question is "does it help for _____?" Most things you put in that blank the answer will be "no".
6
0
40
@techleadhd
You think coding is a "brutal" job, which makes you easily the weakest person in this discussion.
3
0
38
@lillooeterin
I live in the Vancouver of the Southern Hemisphere. I see sunshine in 6 minutes micro doses.
4
0
38
@Abebab
Zero shot inference means unknown performance and therefore not fit for production in anything that matters.
3
2
37
This is genuinely the dumbest thing I have ever seen.
Last night, no one at
@MIT
had a good night’s sleep.
Yesterday evening, shortly after I posted that we were launching a plagiarism review of all current MIT faculty, President Kornbluth, members of MIT’s administration, and its board, I am sure that an audible collective gasp
9K
9K
52K
4
3
36
@misha_saul
@clairlemon
So you didn’t read the output from Claude and do the most basic research?
1
0
33
@ElizAthertonSop
@sdbaral
@lpachter
@EricTopol
@WesPegden
@VPrasadMDMPH
Acting fast has always been a better response to this virus than writing new installments to the Great Barrington Declarations.
6
4
35
I’m a bit late to this conclusion, but there genuinely isn’t a better model for grounded RAG than
@cohere
Command-R+
1
5
37
It’s basically impossible to follow AI people on this website and not have my feed clogged with nauseating tweets like this.
3
1
33
@KanakaMPH
It's bad for calcium homeostasis (and by extension bone health) to have low levels. The threshold for "low" is probably set too high by most labs/doctors and falsely classifies most people as deplete. That's about the extent of it.
1
0
33
@timgill924
They've got a philosophy of everything. Philosophy of Science? If you're so obsessed with science, DO science. Those who can, do; those who can't, write philosophy about it.
6
1
33
@GrogsGamut
See, even economists don’t trust their crappy models (the main implication being that economists should be the first to lose their jobs given how little value they add)
1
1
33
@ylecun
@kind_inkind
@davidmj
@GaryMarcus
You need to stop bringing up your open access publications and open source code. You are an independently wealthy person who does not NEED to make money from your written works. It seems like you haven’t actually thought through the reality of other people’s lives.
2
0
32
Currently enjoying life in the unregulated libertarian AI haven of Australia.
2
2
33
@TracyBethHoeg
Cool, and in Singapore the opposite has happened. Anecdotes are fun, aren't they?
6
0
32
How many of the 38,000 PET scan reports at my hospital were reported to have features of polymyalgia rheumatica? 305. Did I read all 38,000 to find this out? Hell no, I built a model to do it for me.
2
0
32
Our healthcare system is being privatised by stealth and there is no clearer example than my specialty, rheumatology. We're well short of the manpower required to meet patient needs, but the shortfall is significantly greater in public practice.
1
18
32
Controversial take for the tech bros, but obvious one for the clinicians: I don’t see how I would derive any value from GPT-4 or Med-PaLM2 to help with clinical decision making.
13
3
29
This is exactly the sort of unhinged content I like to read from the comfort of my non-blocked alt account.
1
0
29
When is Ioannidis coming out with a preprint on the case fatality rate of tsunamis?
0
4
31
@mmitchell_ai
@timnitGebru
There are so many startups doing this kind of pseudoscience. It's depressing :(
1
4
29
@RahulDBarmanray
But what is "replete"? You could give me 3 different answers to that question and I still wouldn't question your sanity.
2
0
28
A little bit more self-promotion: machine learning to differentiate GCA from its mimics
#ACR21
4
11
29
@joshgnosis
Authoritatively Writing About Things I Don't Understand, The Adam Creighton Story
2
1
27
Two utterly broken modes of thinking have been revealed by this pandemic: induction-only RCT-obsessed EBMers (think Vinay Prasad, GBD) and the extreme reductionists who cannot fathom that most basic science fails in vivo (think ivermectin-heads/crypto bros).
3
3
27
🚨Non-COVID Preprint🚨
We've been working on the problem of Adverse Drug Reaction (ADR) detection using machine learning at the Medicines Optimisation Service
@Austin_Health
@Austin_HealthRx
2
8
24
@HeidyKhlaaf
“It’s just like a human” is basically a confession that the person hasn’t interacted with many humans.
0
0
26
@delaBJL
@hadleywickham
@djnavarro
"I'd like to purchase the domain "
" is not available"
"I know what it stands for, but is the domain available?"
Data science gets its own Who's on First?
0
1
27
@ylecun
@kind_inkind
@davidmj
@GaryMarcus
Now open source the entirety of Facebook and all of its collected data.
1
0
26
@HillCountryDiz
@kevinnbass
The simple answer: being deplete is bad for bone health and most claims beyond that are the closer to marketing than science.
5
1
25
Cytokine I dislike: TNF-α
Cytokine I begrudgingly respect: IL-1
Cytokine I think is overrated: IL-6
Cytokine I think is underrated: IL-10
Cytokine I like: IFN-γ
Cytokine I love: IL-2
Cytokine I dream of blocking: IL-8
8
3
26
Love to see a CEO, with no medical training, give specific medical advice backed by zero evidence, whichh if followed would benefit him financially. Deranged stuff going on in this thread.
@mikellpzdetorre
@ezrainc
We recommend that every adult gets at least a baseline scan, and then annual scans > 40.
MRIs don’t use ionizing radiation, so they are safe to do.
59
3
6
0
9
26
@GidMK
Oh, just wait until you read the evidence on breast feeding, tongue tie, infant sleep, milestones… it’s a house of cards populated by gurus.
0
0
25
@RobertdeGraaf2
@bryancsk
This conversation is tedious, but hopefully instructive for non-Australians who didn’t realise just how lacking in imagination we are.
1
0
22
@RealGatorHours
@codingbrown
Simultaneous occurrence in both legs is extremely rare. Clinicians consistently overestimate compound probability calculations
This just out: Physicians do not logically/correctly estimate the probability of outcomes resulting from sequences of events - a thread.
#conjunctionfallacy
#probability
#medicaldecisionmaking
#clinicalreasoning
#innumeracy
#numeracy
@JAMANetworkOpen
16
80
240
0
0
23
@abacaj
“Hand derived back prop steps - let our artisans design the perfect gradient for your needs”
0
1
23
@ChristosArgyrop
The pro-virus nuts are going to have a helluva time reconciling the RSV surge in Sweden with their ideas about Sweden not having a lockdown and post-lockdown causing RSV surges.
3
3
23
Thanks to
@explosion_ai
for highlighting our work. Lots more to come in improving ADR reporting for medication safety
@DrDouglasPires
@drdavidliew
@wendywchapman
@LizSu_MOS
Natural Language Processing for Adverse Drug Reaction (ADR) Detection (by
@Austin_Health
,
@DrCMcMaster
)
The project has fine-tuned a NER model from the
@huggingface
model hub using a JSONL file of annotations exported from
3
34
128
0
6
23
@Michigan_Noah
@MarkELindsay
Vinay only adopted this perspective once the cabal of weird JPI-associated journals started publishing all of his opinions with zero editorial oversight.
1
0
20
EBV playing 5D chess while vitamin D is playing checkers.
0
1
21
@Molson_Hart
@TuanPham672604
@ClementDelangue
@ylecun
@elonmusk
Do not let your Google search fool you, you cannot replace the CNNs in your phone with any of those architectures.
0
0
20
@psufka
@perplexity_ai
@YouSearchEngine
is definitely better and more flexible than perplexity. Would strongly recommend trying it out. The CEO
@RichardSocher
is a veteran AI/NLPer and it shows in the attention to detail.
1
1
21
@solisSATX
💯 With absolutely everything that these grifting academic institutions use to suck us dry.
0
0
20
If you post incessantly about elites not understanding people with low IQ, you are one of those people. Intelligent people who interact with the breadth of society on a daily basis do not post like this, because they understand that it’s a gross simplification.
Most people who will read this tweet cannot fathom what 80 IQ really means
If you can easily understand Jane Austen, complete this math problem, and parse logical statements with multiple conditionals, you’re likely 115+
People at or below median can’t even do algebra
198
55
1K
1
3
21
I don't have anything to promote other than ideas. But if you liked this idea, you might like other centrally-planned things like healthcare or the economy.
1
1
21
@DrJBhattacharya
I admire your consistency, Jay. Must take some effort to be so consistently wrong.
0
0
20
@DrLiver
Oh, don't be so cynical. He's just warning parents about the really serious side effects of n95 use, like [redacted] and [insert insane things from your imagination] and [hey, I'm not responsible for spreading lies if I just trail off and invite wild speculation...]
0
0
19
If this was 2015 we would all be sharing our results from the buzzfeed "which COVID variant are you" quiz.
3
1
19