#AI Medical

[quadranthealth]

AI Medical Scribes Are Changing The Game

An AI medical scribe is a secure computer software that listens to your physician-patient conversations, transcribes it, and, typically, packages it into an EHR-ready clinic note. They can be used to automate documentation and clinical note taking, allowing physicians more time to be more present during patient interactions. 

[reasonsforhope]

I'm worried about the rising rate of young adults getting cancer.

For what it's worth, we've actually made a shocking amount of progress against cancer - especially the most common cancers like breast cancer, and especially in the past 30 years.

Cancer rates have been falling, often dramatically (x, x, x, x, x, x). One of the best examples it that breast cancer deaths in the United States dropped 58% between 1975 and 2019 (x).

Right now, we're at the beginning of an absolute revolution in cancer care that promises to increase survival rates even further. This revolution has been going on to a lesser degree since the first human genome was successfully sequenced in the early 2000s (and in fact, the first gapless sequencing of a human genome was finally finished just two years ago, in 2022), and to a greater extent since CRISPR DNA-editing technology was first successfully tested in 2013, and since medical digitzation/digital communication and vaccination were massively spurred ahead in 2020, by the COVID pandemic (x, x).

Right now, the results of this revolution are only beginning to trickle out into actual treatments. But I guarantee you, in the next one to three decades, the way we fight cancer will be massively transformed.

We're talking personalized genome sequencing for each person with cancer - not just for early and better detection, but even to figure out what types of treatments will work best. (x, x, x, x)

We're talking using CRISPR-based DNA editing to literally cut cancer-causing mutations out of your DNA, to edit the genes of immune cells to better detect and kill cancer cells, and to kill cancer-causing viruses. (x, x, x, x)

We're talking using CRISPR-based screening to figure out how chemotherapy resistance works, so that we can overcome it - and even weaponize it. (x, x)

We're talking using CRISPR to edit immune cells so that they recognize and target the mutations of a single individual's specific tumor. (x)

We're talking new types of testing that can predict if cancer will return years before it shows up on scans. (x)

We're talking using (non-generative) AI to massively increase the accuracy and earliness of cancer detection - which by the way is already starting to happen, there are several AI-based systems that detect cancer earlier and more accurately than doctors do. (x, x, x, x, x, x)

Also, the more we transition to a green, sustainable, and ethical future, the fewer cancer-causing substances will be in the environment (fossil fuels, oil drilling, and mining are massive sources of carcinogens at every point in the process).

Cancer is awful. That is a massive understatement. But the fight against cancer is one where there are so many reasons for hope.

[godbirdart]

i've been so tired of google's new legal liability bot sitting on top of the existing highlighted result, taking up page space and either parroting exactly what the highlighted result said or offering completely unrelated or incorrect results, that i actually cheered when this came up:

"oh but AI is experimental -"

the bot told people to eat glue on pizza. you can talk yourself blue in the face about the bot's learning curve and how "it'll be improved with time", but maybe a bot being touted as the latest and greatest in scouring the internet for accurate information should not come with a permanent glaring disclaimer of "it's still learning / results may not be accurate!" as the generative AI ouroboros continues to keep on chewing.

[zuzsenpai]

I feel like people have been warning about this and I completely ignored it. Deep sigh as I force myself to remember to add “-ai” at the end of every fucking search now

(Is there an extension that turns this off??)

[thegodemperorsmycopilot]

Sister Elysia, Hospitaler

A friend's character from our group's Wrath & Glory campaign.

[mostlysignssomeportents]

Privacy first

The internet is embroiled in a vicious polycrisis: child safety, surveillance, discrimination, disinformation, polarization, monopoly, journalism collapse – not only have we failed to agree on what to do about these, there's not even a consensus that all of these are problems.

But in a new whitepaper, my EFF colleagues Corynne McSherry, Mario Trujillo, Cindy Cohn and Thorin Klosowski advance an exciting proposal that slices cleanly through this Gordian knot, which they call "Privacy First":

https://www.eff.org/wp/privacy-first-better-way-address-online-harms

Here's the "Privacy First" pitch: whatever is going on with all of the problems of the internet, all of these problems are made worse by commercial surveillance.

Worried your kid is being made miserable through targeted ads? No surveillance, no targeting.

Worried your uncle was turned into a Qanon by targeted disinformation? No surveillance, no targeting. Worried that racialized people are being targeted for discriminatory hiring or lending by algorithms? No surveillance, no targeting.

Worried that nation-state actors are exploiting surveillance data to attack elections, politicians, or civil servants? No surveillance, no surveillance data.

Worried that AI is being trained on your personal data? No surveillance, no training data.

Worried that the news is being killed by monopolists who exploit the advantage conferred by surveillance ads to cream 51% off every ad-dollar? No surveillance, no surveillance ads.

Worried that social media giants maintain their monopolies by filling up commercial moats with surveillance data? No surveillance, no surveillance moat.

The fact that commercial surveillance hurts so many groups of people in so many ways is terrible, of course, but it's also an amazing opportunity. Thus far, the individual constituencies for, say, saving the news or protecting kids have not been sufficient to change the way these big platforms work. But when you add up all the groups whose most urgent cause would be significantly improved by comprehensive federal privacy law, vigorously enforced, you get an unstoppable coalition.

America is decades behind on privacy. The last really big, broadly applicable privacy law we passed was a law banning video-store clerks from leaking your porn-rental habits to the press (Congress was worried about their own rental histories after a Supreme Court nominee's movie habits were published in the Washington City Paper):

https://en.wikipedia.org/wiki/Video_Privacy_Protection_Act

In the decades since, we've gotten laws that poke around the edges of privacy, like HIPAA (for health) and COPPA (data on under-13s). Both laws are riddled with loopholes and neither is vigorously enforced:

https://pluralistic.net/2023/04/09/how-to-make-a-child-safe-tiktok/

Privacy First starts with the idea of passing a fit-for-purpose, 21st century privacy law with real enforcement teeth (a private right of action, which lets contingency lawyers sue on your behalf for a share of the winnings):

https://www.eff.org/deeplinks/2022/07/americans-deserve-more-current-american-data-privacy-protection-act

Here's what should be in that law:

A ban on surveillance advertising:

https://www.eff.org/deeplinks/2022/03/ban-online-behavioral-advertising

Data minimization: a prohibition on collecting or processing your data beyond what is strictly necessary to deliver the service you're seeking.

Strong opt-in: None of the consent theater click-throughs we suffer through today. If you don't give informed, voluntary, specific opt-in consent, the service can't collect your data. Ignoring a cookie click-through is not consent, so you can just bypass popups and know you won't be spied on.

No preemption. The commercial surveillance industry hates strong state privacy laws like the Illinois biometrics law, and they are hoping that a federal law will pre-empt all those state laws. Federal privacy law should be the floor on privacy nationwide – not the ceiling:

https://www.eff.org/deeplinks/2022/07/federal-preemption-state-privacy-law-hurts-everyone

No arbitration. Your right to sue for violations of your privacy shouldn't be waivable in a clickthrough agreement:

https://www.eff.org/deeplinks/2022/04/stop-forced-arbitration-data-privacy-legislation

No "pay for privacy." Privacy is not a luxury good. Everyone deserves privacy, and the people who can least afford to buy private alternatives are most vulnerable to privacy abuses:

https://www.eff.org/deeplinks/2020/10/why-getting-paid-your-data-bad-deal

No tricks. Getting "consent" with confusing UIs and tiny fine print doesn't count:

https://www.eff.org/deeplinks/2019/02/designing-welcome-mats-invite-user-privacy-0

A Privacy First approach doesn't merely help all the people harmed by surveillance, it also prevents the collateral damage that today's leading proposals create. For example, laws requiring services to force their users to prove their age ("to protect the kids") are a privacy nightmare. They're also unconstitutional and keep getting struck down.

A better way to improve the kid safety of the internet is to ban surveillance. A surveillance ban doesn't have the foreseeable abuses of a law like KOSA (the Kids Online Safety Act), like bans on information about trans healthcare, medication abortions, or banned books:

https://www.eff.org/deeplinks/2023/05/kids-online-safety-act-still-huge-danger-our-rights-online

When it comes to the news, banning surveillance advertising would pave the way for a shift to contextual ads (ads based on what you're looking at, not who you are). That switch would change the balance of power between news organizations and tech platforms – no media company will ever know as much about their readers as Google or Facebook do, but no tech company will ever know as much about a news outlet's content as the publisher does:

https://www.eff.org/deeplinks/2023/05/save-news-we-must-ban-surveillance-advertising

This is a much better approach than the profit-sharing arrangements that are being trialed in Australia, Canada and France (these are sometimes called "News Bargaining Codes" or "Link Taxes"). Funding the news by guaranteeing it a share of Big Tech's profits makes the news into partisans for that profit – not the Big Tech watchdogs we need them to be. When Torstar, Canada's largest news publisher, struck a profit-sharing deal with Google, they killed their longrunning, excellent investigative "Defanging Big Tech" series.

A privacy law would also protect access to healthcare, especially in the post-Roe era, when Big Tech surveillance data is being used to target people who visit abortion clinics or secure medication abortions. It would end the practice of employers forcing workers to wear health-monitoring gadget. This is characterized as a "voluntary" way to get a "discount" on health insurance – but in practice, it's a way of punishing workers who refuse to let their bosses know about their sleep, fertility, and movements.

A privacy law would protect marginalized people from all kinds of digital discrimination, from unfair hiring to unfair lending to unfair renting. The commercial surveillance industry shovels endless quantities of our personal information into the furnaces that fuel these practices. A privacy law shuts off the fuel supply:

https://www.eff.org/deeplinks/2023/04/digital-privacy-legislation-civil-rights-legislation

There are plenty of ways that AI will make our lives worse, but copyright won't fix it. For issues of labor exploitation (especially by creative workers), the answer lies in labor law:

https://pluralistic.net/2023/10/01/how-the-writers-guild-sunk-ais-ship/

And for many of AI's other harms, a muscular privacy law would starve AI of some of its most potentially toxic training data:

https://www.businessinsider.com/tech-updated-terms-to-use-customer-data-to-train-ai-2023-9

Meanwhile, if you're worried about foreign governments targeting Americans – officials, military, or just plain folks – a privacy law would cut off one of their most prolific and damaging source of information. All those lawmakers trying to ban Tiktok because it's a surveillance tool? What about banning surveillance, instead?

Monopolies and surveillance go together like peanut butter and chocolate. Some of the biggest tech empires were built on mountains of nonconsensually harvested private data – and they use that data to defend their monopolies. Legal privacy guarantees are a necessary precursor to data portability and interoperability:

https://www.eff.org/wp/interoperability-and-privacy

Once we are guaranteed a right to privacy, lawmakers and regulators can order tech giants to tear down their walled gardens, rather than relying on tech companies to (selectively) defend our privacy:

https://pluralistic.net/2022/11/14/luxury-surveillance/#liar-liar

The point here isn't that privacy fixes all the internet's woes. The policy is "privacy first," not "just privacy." When it comes to making a new, good internet, there's plenty of room for labor law, civil rights legislation, antitrust, and other legal regimes. But privacy has the biggest constituency, gets us the most bang for the buck, and has the fewest harmful side-effects. It's a policy we can all agree on, even if we don't agree on much else. It's a coalition in potentia that would be unstoppable in reality. Privacy first! Then – everything else!

If you'd like an essay-formatted version of this post to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2023/12/06/privacy-first/#but-not-just-privacy

Image: Cryteria (modified) https://commons.wikimedia.org/wiki/File:HAL9000.svg

CC BY 3.0 https://creativecommons.org/licenses/by/3.0/deed.en

[crewtawn]

If you wanna use any of my art pleeease credit me pleaseeee...

[bibbysstuff]

What would happened if we get sick? What would kinito do?

nurse you back to health of course!

[gayestcowboy]

one of the reasons i hate generative AI is that i have a chronic injury that makes me unable to write by hand for longer than a few sentences and therefore i need a keyboard in order to write essays for classes, and i’m a bit worried that professors won’t allow me to type essays as a result of potential AI use, even though i have a perfectly valid medical reason to need a keyboard. i’ve seen a few posts online saying that the only way to prevent students from relying on AI is to make them write by hand, and while i understand the sentiment and don’t even necessarily disagree with the general point, the matter of accessibility still needs to be addressed somehow, and it’s just a bit frustrating on a personal level as someone who has had to fight with teachers and professors in order for my needs to be met even before the era of chatgpt

[endivinity]

Would you be ok with indivituals using some of your fallouty unique deathclaws and other creature drawings as references when theyre running a private Fallout TTRPG campaign amongst friends?

Totally ok if youre not fine with that! I just wanted to check.

it may be possibly time for another long-winded explanation that will enrage precisely two people who will send me weird anon hate over it because they don't understand the notion of transformative works but - It's hard to state in plain terms what I am and aren't comfortable with people using my art for, because even for me there's a lot of handwavey 'am i really uncomfortable with this or is it just a kneejerk reaction' kind of thing with a lot of caveats

there's often a pretty big disconnect between what people view as references - some people will view it as a single image, like "for reference, this is what it looks like" and slap the art down on the table for their players to go WHOAAAA over (repost blogs tend to do this). For an artist doing this, it's usually as a study, or results in a pretty obvious 1:1 because the idea they're pulling from isn't diluted across a range of things.

The other use of reference is several images and enough of your own personal touch to put a unique spin on all of them in a cohesive artwork or design. This is the bit that a lot of people stumble over, because they go 'but I like this thing only' and don't want to venture beyond that. If you're able to not only pull a range of artworks, but a range of artworks by different artists, immense kudos to you. The way I construct deathclaws is from an immense knowledge of weird animals and different media. For instance, Spectral makes use of a deathclaw, a ghost leviathan from subnautica, a xenomorph, and the understanding of vestigial limbs, bioluminescence, diaphanized tissue, and opalization. Transforming this in a tabletop might therefore look like the image itself, but then adding say the aquatic spinosaurus theory in there and making it swim out of an irradiated lake with a paddle tail and a huge back crest. And a bigger mouth with worse teeth that can strip a human's arm down to the bone in one degloving bite. You don't necessarily have to be able to draw it (I'm in a bit of a niche and therefore shouldn't hold everyone to my same standards) but it also says good things about your ability to host a tabletop game if you're able to be creative with the unique ideas you put in it and your ability to visualize and describe them. Embrace that. (Some of the kickback against this was people going 'well artists do this all the time, they take other artists' works for their references, that's part of the industry standard' but that's the point of transformative works and not typically modern tabletop gaming. I'm also just one person doing this. I'm not an industry professional, I'm not a huge company for which my works are publicly available in an immensely popular IP. It may be fanart and I don't own deathclaws, but I still own all rights to the art itself. Some people (that one really furiously angry anon in particular) hold me to the same standard as if I was representative of Bethesda Softworks itself and therefore it's right and proper to take my shit, because it's deathclaws, and all deathclaws are Bethesda's, and I wouldn't be this popular without that, I should expect people to take my stuff, it's the internet - I am just one person making fanart. and I am very tired.) I think the biggest problem I have with people taking my designs for TTRPG assets is that it's the only reaction they have sometimes. the 'wow cool! can I take this?' reaction akin to a little child shoving things in their mouth. That doesn't reflect well on you, and for the artist it doesn't feel good. And most tabletop gaming these days is casual sessions that usually center around getting the campaign itself done with little creativity beyond what the players bring to the table, which results in using other peoples' art they found on google without being creative about it at all, which is why you'll see a lot of artists who have beef with it, because it also doesn't feel good. All this to say - if you ask and are respectful and credit back, it still feels weird to me, but like... sure! I do this for fun and to express creativity, so if it encourages other people to also have fun and express their creativity, I'd feel bad saying no to an earnest request! It's fanart, we're all fans here, etc But also most people who aren't respectful won't ask, won't be creative, or will get mad when I say Can You Don't, so I'm kinda preaching to the wrong crowd here - to those people, you're right, I can't stop you from doing it anyway. but it will not put you in my good books

[reasonsforhope]

"An international research team has found almost a million potential sources of antibiotics in the natural world.

Research published in the journal Cell by a team including Queensland University of Technology (QUT) computational biologist Associate Professor Luis Pedro Coelho has used machine learning to identify 863,498 promising antimicrobial peptides -- small molecules that can kill or inhibit the growth of infectious microbes.

The findings of the study come with a renewed global focus on combatting antimicrobial resistance (AMR) as humanity contends with the growing number of superbugs resistant to current drugs.

"There is an urgent need for new methods for antibiotic discovery," Professor Coelho, a researcher at the QUT Centre for Microbiome Research, said. The centre studies the structure and function of microbial communities from around the globe.

"It is one of the top public health threats, killing 1.27 million people each year." ...

"Using artificial intelligence to understand and harness the power of the global microbiome will hopefully drive innovative research for better public health outcomes," he said.

The team verified the machine predictions by testing 100 laboratory-made peptides against clinically significant pathogens. They found 79 disrupted bacterial membranes and 63 specifically targeted antibiotic-resistant bacteria such as Staphylococcus aureus and Escherichia coli.

"Moreover, some peptides helped to eliminate infections in mice; two in particular reduced bacteria by up to four orders of magnitude," Professor Coelho said.

In a preclinical model, tested on infected mice, treatment with these peptides produced results similar to the effects of polymyxin B -- a commercially available antibiotic which is used to treat meningitis, pneumonia, sepsis and urinary tract infections.

More than 60,000 metagenomes (a collection of genomes within a specific environment), which together contained the genetic makeup of over one million organisms, were analysed to get these results. They came from sources across the globe including marine and soil environments, and human and animal guts.

The resulting AMPSphere -- a comprehensive database comprising these novel peptides -- has been published as a publicly available, open-access resource for new antibiotic discovery.

[Note: !!! Love it. Open access research databases my beloved.]"

-via Science Daily, June 5, 2024

[benjaminthewolfcasual]

This Is Now Canon And Nobody Can Tell Me Otherwise

(This is Medic on Character.ai)

I NEED someone to draw this now- /nf

[aivoluptulicious]

An experienced pharmacy clerk, feeling hot in the aisle. Doubling down on medicine?

[dryndelicate]

When you are done straightening your stockings, Nurse Angel, please rush back to our VIP patient, Gigi Princeton, in room 8. Monitoring her recovery is extremely important, especially on a Sunday, when the doctors are out.

Face swap courtesy of @angelmiledg. Get well, @gigiprinceton!

[robogamer360]