The drug company will credit key steps of the discovery or steps in the treatment process to AI/ML techniques. The treatment must be approved before 2030 by the FDA for a condition afflicting at least 1 million people at the time, which had no previous cure, rarely goes into remission, and without treatment either significantly shortens lifespan or is considered extremely debilitating. Treatments may exist at the time, but this will be a complete cure with an efficacy of >50%. The treatment may need to be repeated if the illness recurs, but isn't needed on a continuing basis.
Examples (not limited to):
An incurable virus, bacterium, or parasite.
Degenerative illnesses such as Alzheimer's, diabetes, or cardiovascular.
Physical such as paralysis, amputation, and many types of blindness or deafness.
Mental health disorders such as autism, bipolar disorder, and schizophrenia.
This is way too high. To be approved by the end of the decade the molecule needs to be discovered already. Trials take forever.
@LachlanMunro also base rates of cures meeting this criteria are pretty low.
ai has been used in huge amounts of drug discovery for several years. there's a pretty good chance a big deal molecule has been discovered already; though, I have not at time of comment done the research to find a specific example.
I'm torn because
It's not often an outright "cure" gets invented
AI/ML is starting to be incorporated into everything in one way or anyone, so if only for hype reasons I expect drug companies to claim that it played a major role if any such cure is invented
@JonathanSimon so really I see this more of a bet on "do drug companies cure a major endurable illness by 2030".
@JonathanSimon excellent edge case!
The answer - prevention differs from a cure. It's possible a treatment could serve as both a vaccine and a treatment, but it only counts for the purposes of resolving this market if it cures people who already have the illness. If they have an illness, even without symptoms, and it significantly shortens lifespan, and the treatment resolves the issue permanently (for >50% of people treated) then it counts. If the treatment must continue for life, even if the vaccine aspect is permanent, then it wouldn't count.
@JonathanSimon If I don't believe that AI actually was instrumental in the design, manufacture, or treatment then I won't resolve to a yes. As part of the FDA process they'll have to share more details.
There's also a chance that drug companies won't want to share the techniques they used or that they won't want to risk negativity from the public, and will try to hide the use.
@RealityQuotient Thank you for the clarification! (And sorry for the typos... really wish there was an edit button)
In that case I think this market is pretty overconfident on YES, in large part because of how infrequently we produce outright cures for serious long-standing diseases, but also because the biggest recent AI breakthroughs haven't been in realms that would be game-changers for drug development. And of course the FDA being notoriously slow at approving anything substantially new.
@JonathanSimon You're correct. Cures are very, very rare. People signficiantly overestimate the number of illnesses that can be cured through an FDA approved treatment. The bar is also high on this market due to the fact 1M people must be afflicted for the cure to resolve the market. Surgery, unlike non-invasive medicine, often cures a condition (e.g. any organ transplant), and there are surgical robots. That said, the FDA doesn't have oversight over surgical procedures. Now, if an autonomous surgical robot had a program to perform a surgery autonomously, would the FDA step in to regulate that?
A high potential area where cures may occur in this decade is in oncology. Cancer isn't effectively treated without curing it. It's also a space that's regulated by the FDA. There has been some recent success in base editing, and it's likely that AI would be needed for the process of generating the personalized cure.
With antibiotic resistance on the rise, another promising area would be an FDA approved macrophage based treatment that selects a macrophage to treat infections using an AI pairing tool. Right now it's a manual process and is slow and doesn't scale well - but imagine every hospital with a room full of macrophages ready to be deployed based on a test result. I'm not in medicine so I don't understand the barriers to general availability here, but it's a possibility.
While cures are rare, we're still at the beginning of the bathtub curve of what AI can do. That is, we may find many directions that are hard for humans and easy for AI to help with.
@RealityQuotient I can see that you've thought about this a lot! I agree that curing some specific type of cancer is one of the more likely ways for this market to resolve positively. Surgical robots are an interesting one that hadn't occurred to me. I would guess that an autonomous surgical robot would be classified as a "medical device" and thus regulated by the FDA. But again, robotics is an area that's seen very little substantial progress from AI breakthroughs in the past decade, as highlighted by OpenAI scrapping their robotics division.
@RealityQuotient do you mean bacteriophage?
With markets like this you vote no and then it is a win-win. 😎
@ZiquaftyNny why? (Honest question, I'm still trying to understand this platform)
@JonathanSimon In the real world you want it to be yes but if you vote no on the platform then at least you make some fake money if the future is not good.
Full Discloser: I have not invested in this market at all.
p.s sorry for late response.