[core-questions]

> In the future, the Cleveland Clinic will also purchase a 1,000+ qubit IBM next-generation quantum system for installation at a data center in its home town.

LOL, they better actually hit some of their targets to get past 53 qubits if they want to live up to this bullshit hype.

For all IBM has been talking about their QC work, they still only have a measly handful of noisy useless qubits that can't do anything worth writing home about. It would be hard to believe that a medical company would want one, until we remember all the back-room deals and nepotism inherent to the American medical industry. Years' of people's insurance payments frittered away on a magical machine that won't do anything to help anyone's medical outcomes compared to spending the same amount of money telling Yanks to stop shoving fried sugar into every hole.

[gus_massa]

It looks worse than the "applications" of Watson to medicine.

[core-questions]

I'm actually totally willing to believe that ML for medicine is a great application. When I go to a doctor for a non-emergency issue, what am I doing other than describing a bunch of symptoms, and then the doctor looks within their memory to see if it is something they know about, or if they need more information to be sure?

Hell, my doctor fires up some kind of Doctor Google, proprietary paid system, and searches for things all the time. It doesn't even offend me, damn knows that's a huge component of my job as well. Extending that to inputting a formal description of my symptoms and narrowing down the results doesn't seem like a bad move.

Even better would be tying something like HealthKit or other data sources into the system. Need to make a food log or a bowel log or whatever? Why _not_ use an app and actually get structured data some system can make sense of?

Here's the thing: gate-model QPUs are a decade or more off from being applicable to this problem. Quantum annealers, on the other hand, are basically neural-network machines in a box, so if you want to apply quantum computing to this problem, IBM probably isn't the vendor to go with. Their QAOA work doesn't look like it's any better than using an actual annealer, to boot.