Didn’t know there was a market for coding session watchers. Let me know if someone’s interested in watch me slow-code my current long running side project: Focusly.
Unless the AI can talk tot the patient to get some context, it's going to be taking decisions with only very partial information, no matter how good it is.
Still, I'm thinking that as it improves, it's going to show that doctors are not that good at their job on average, and that's going to be fun to watch.
> Still, I'm thinking that as it improves, it's going to show that doctors are not that good at their job on average, and that's going to be fun to watch.
Medical AI is trained on labels generated by doctors. Can you explain how it will exceed the performance of doctors on average? Are you assuming that the labels will be generated by the "top x%" of doctors? If so, how will you identify those individuals? Or is there some other mechanism you're expecting to improve the performance?
>I think AI should always be a support tool, but people are always talking about replacing doctors...
Because Americans are trying to solve their healthcare crisis by attacking the doctors and literally not the entire rest of the industry which controls the spiraling healthcare costs.
The end of result if we manage to eliminate doctors is still spiraling out of control healthcare.
Radiologists aren't talking to the patient either and don't have earlier scans available, they "just" comment on the image; so the doctor who does have the patient, their history and the radiology report(s) might as well receive the report from a ML system instead of a human radiologist.
That's basically how most modern frameworks do it.
Django-Rest-Framework does that with view + serializer + validators.
A more recent example, with a leaner and cooler implementation is FastAPI, where type hints are also use to declare validation on your end point, while a model is used for serializing the response.
A bit of a plug, but if you are using python in your project already, give a try to pydoit (pydoit.org/). It completely replaced make, task runners or build systems for me.
Pydoit is the sweet spot for most of my use cases:
- it scales up (graph of deps, file watcher, etc) but above all, it scales DOWN (simple stuff are dead simple)
- it promotes a declarative task definition
- it embraces the shell yet let you use python if you need to
- it gets out of your way and doesn't try to rule your project. It just runs what you want and disapear.
- the doc is nice
The biggest drawback is that you need to pip install it, which means non python devs will avoid it. I wish it was provided as an stand alone executable
pydoit is not single language in the sense it limits your to tasks for your language. It's completely agnostic in that sense, and I use it for non python projects as well.
Even the fact it's using python to declare your tasks is not a problem: it's basically a function signature and a mapping, nothing you can't master in 5 minutes, and certainly simpler than make files DSL. Also easier to get right and debug.
The problem is the fact you need python to install it and run it, which non python dev will rightfully not care to do.
If you use react or vue, you can plug components precisely anywhere on the page, you don't need to take over the page entirely.
In fact, with vue, you can even put the template in the HTML directly, and just hook the logic on to it, angularjs v1 style.