Alan

intelligence artificial - decoding bone marrow to improve human health -

[Alan @alanaberdeen]

@pkoo562 Sadly also 1) has a viral nature to it

Reviewers of future papers asking authors to demonstrate how they compare on the misaligned benchmark forevermore.

[Alan @alanaberdeen]

More on MegaCALR: AI-Driven, Single-Cell Quantification of CALRmut in Megakaryocytes:

https://t.co/CtQrdkQRVA

[Alan @alanaberdeen]

Two cool things from us at EHA:

1. Collab with Incyte: AI analysis of fibrosis and megakaryocytes characterised patients and correlated with outcomes in the MOST study

2. AI-driven quantitative characterisation of CALRmut IHC -- timely for given the rise of targeted therapies

[Alan @alanaberdeen]

More on AI-Quantified Bone Marrow Biomarkers in the MOST MPN Study here:

https://t.co/PhQAsuORQL

[Bo Wang @BoWang87]

Yann LeCun just said something that every AI-in-healthcare researcher should sit with.

He basically said:
If language were enough to understand the world, you could learn medicine by reading books.

But you can’t.
You need residency. You need to see thousands of normal cases before you recognize the abnormal one.
He also points out something wild — all the public text on the internet is on the order of 10¹⁴ bytes.

A 4-year-old processes about that much through vision alone.
The world is just… higher bandwidth than text.

I think this shift — from language models to world models — is going to matter a lot in healthcare. 🫀

[Ron Alfa @Ronalfa]

Yes, and especially in BIO, the right data is not just an advantage, it is everything.

It has been challenging to do FM model partnerships with pharma, because they have no use for models trained on your bespoke experiment.

However, everyone values FMs trained on human tumors from the clinic to improve their oncology clinical pipeline.

[Alan @alanaberdeen]

@Ronalfa This.

Have a feeling the incentives around publishing pushes people toward architecture tweaks when they’d get further just improving their data.

[Alan @alanaberdeen]

Either way; isEven in the mud

[Alan @alanaberdeen]

All those tiny npm packages are increasingly pointless. AI-generated code makes it trivial to write your own little utility instead of pulling in a dependency with its own unknowns

[Alan @alanaberdeen]

Though those packages are probably a big chunk of training data. Now we get stochastic copy-paste instead of npm install

[derek guy @dieworkwear]

I can tell you what's stopping most men from dressing like this. Please click open this thread (into a new window) so that you can see the correct placement of my photos. This will help you understand my argument more clearly.

We should first identify what we're looking at. This is James Stewart in the 1946 film It's a Wonderful Life, now beloved as a holiday classic, but during the immediate postwar years, it was eyed with suspicion. During the early days of the Cold War, the FBI thought this film was communist propaganda. After all, the film is about a kind-hearted man who sacrifices his own self-interest to help others. He starts an organization to help working-class families secure affordable housing and takes a stand against the villainous banker Mr. Potter. The FBI felt this was an attack on the upper classes, especially since two of the screenwriters were suspected of being communist sympathizers, so they referred the film to the House Un-American Activities Committee, although no action was ultimately taken.

For the menswear-minded, the film will stand out especially. Despite portraying the humble George Bailey, James Stewart's wardrobe is remarkably good in this film. We see him in some things that today might be judged as strange — such as a peak-lapel single-breasted tweed suit with four (four!) patch pockets, including two at the breast — but for the most part, his clothes have aged very well. Meaning, you could take many of these clothes and wear them today (hence the original poster's question, "what's stopping you from doing so?").

The secret to this success is in the quality of the make and proportions. In the film, Stewart wears trousers that are high enough to cover his shirt when the coat is fastened. The coat bisects him halfway from the collar to the floor. The trousers are full enough to create a smooth silhouette between the bottom and top halves of his outfit. A sharp eye will also notice that the lapel has a very pleasing roll — evidence of hand pad stitching. And of course, the collar always hugs his neck.

The overcoat, which sparked this discussion, is also very long (long enough to reach below his knees, which is what you want in blustery weather). Like the lapels on his suit jacket, the his overcoat lapels are neither overly skimpy nor overly wide — just classic enough so you can never peg the garment to a specific fashion trend or decade.

Given what we know about how actors dressed for films during this era, along with what we can see in photos, we can reliably guess that these garments are both bespoke and hand-tailored. By bespoke, I mean the garments were made from scratch and perfected through three fittings.

The advent of ready-to-wear manufacturing in the mid-19th century, along with the explosion of designer clothing and sportswear in the postwar period, effectively swept away our domestic bespoke tailoring trade. Instead, what's left across the country are primarily made-to-measure shops run by businesspeople, not tailors, who have relationships with overseas factories. This system can be fine, but it may not achieve some of the effects you see here. This is especially true if the shop has been influenced by fashion trends (which most have). Such shops produce shorter, tighter garments made from fine, silky materials that don't achieve this look.

Thus, the simple answer to your question is: most men don't dress like this because they don't have access to bespoke tailors.

However, some men have access to bespoke tailors. In the United States, such people tend to be concentrated in or around major cities, such as New York City or San Francisco. However, even in these areas, the number of bespoke tailors remains small. There are several reasons for this.

First, skyrocketing rents make it very difficult for these businesses to survive. Most people have an upper limit for how much they're willing to pay for the outfit you see in the original image, and that limit is not $10,000. To get bespoke tailoring prices down, we must create affordable housing and commercial real estate.

Second, even if you were to open a bespoke tailoring shop in the US, who would you hire? There aren't many skilled tailors in the US for various reasons. It takes a decade or more to train to be a bespoke tailor — and how would you even do so? Can you survive without health insurance for twenty years? Can you find enough customers to pay for astronomical rents? The US tends to celebrate wealthy entrepreneurs, not craftspeople, and the latter requires slow, steady concentration over decades, often living in poverty until you finally perfect your craft. To help create more craftspeople, we need universal healthcare, affordable housing, and a shift in American values (less worship of money).

Thus, even if you're in NYC or San Francisco, the chances of you getting a bespoke garment from a domestically based tailor are slim. That's why most people who are into bespoke tailoring rely on the many international tailors who swing through major US cities three to four times a year to meet with clients. Such tailors typically hail from the UK, Italy, Japan, or South Korea.

But here we land at yet another problem. During their last tour through the US, the managers behind the South Korean tailoring shop Assisi told me that customs and border agents hassled them, seizing some of their luggage. They consequently lost some of their swatch books, which were critical for their trunk shows. As directed by President Trump, border agents have been unusually harsh to travelers. This creates another barrier for US customers to dress like Jimmy Stewart above.

Let us assume that you're able to see one of these tailors in NYC or San Francisco, and that they've somehow successfully navigated borders and customers without issue. And you're able to repeat this four times — initial meeting to be measured and place an order, then basted fitting, forward fitting, and final fitting. Let's assume there are no other issues, and the item can be shipped to you.

Congrats, you've now been hit with a customs bill. Due to Trump's tariffs, a new tax is levied on incoming shipments (which vary depending on the country of origin). For most bespoke tailors, I've seen this range anywhere from 10% to 20% of the declared value (which can be a lot given the price of bespoke clothes!). If you order something made from cashmere or a cashmere blend, it can be as high as 50% (not an unusual fabric for an overcoat). This creates yet *another* barrier for men to dress like this.

So, to answer the question of why men don't dress like this, a big reason is location. Most men don't live in a major US city. Second, the US doesn't have a culture or climate suitable for raising craftspeople — the worship of money, lack of universal healthcare, and skyrocketing rents (both commercial and residential) make it very difficult to become a bespoke tailor in the US. And if you use one of the international operations, you will have to pray that border agents do not hassle your tailor. And when your garment arrives, you will have to fork over more money to cover Trump's tariffs.

The combination of all these effects makes dressing like this a dream for most men. This is assuming they can get past the cultural stigma of men being less masculine or heterosexual if they express an interest in clothes, which will undoubtedly come through if you dress like James Stewart in the 1940s. Breaking this barrier down requires us to expand our understanding of masculinity or to be less judgmental about gender norms.

[Alan @alanaberdeen]

@brian_lovin Strong agree. Similar to Tufte’s principle - seek to maximise the data-ink ratio.

[Alan @alanaberdeen]

Full publication is Open Access too! https://t.co/wC1bti73Hy

[Alan @alanaberdeen]

Excited to share our latest publication in @Hemasphere_EHA on the impact of AI-driven biomarkers for bone marrow morphology in clinical trials.

Quantitative analysis of bone marrow fibrosis highlights heterogeneity in myelofibrosis and augments histological assessment.

[Alan @alanaberdeen]

This work builds on retrospective analysis of a phase II clinical study of zinpentraxin alfa.

[RC of Pathologists @RCPath]

We are pleased that the Government will take forward the UK National Screening Committee recommendation on use of digital images to identify cancer and speed up diagnosis. #DigitalPathology has the potential to improve #patient care.
https://t.co/lK2Ilqsu3Y

[Alan @alanaberdeen]

Sudden 5xx errors midway through your workflow really highlights your dependency (and vulnerability) on external tools

[Alan @alanaberdeen]

I'd be that OpenAI API downtime is a leading indicator for stars on cool Github repos like this...

[cocktail peanut @cocktailpeanut]

Dalai Alpaca is here!
https://t.co/lm6Kx7QExs

Now you can run the Alpaca LLM on your computer (Mac, Windows, Linux) with just ONE command!

Best of all, all you need is just around 4.2GB of disk space!

Just run this:

[Alan @alanaberdeen]

Here's what GPT-4 thinks of the abstract for the paper that started it all.

[Alan @alanaberdeen]

One of the best uses for LLMs I’ve found so far has got to be Red Teaming - prompt:

“Please provide harsh but fair criticism for a given idea/passage/article/proposal. Layout each of your points with clear concise reasoning and don’t hold back!

[content]”