Engineering major with a bachelors here

OP, have you heard of Computational Biology? It's programming and genetics mixed together, pretty much.

Other than that, I don't really see how Machine Learning could play a role in helping with stuff life cancer diagnosis.

When diagnosing the patient the doctor only knows what he remembers school and from his experiences as a doctor.

A machine could store way more information in it than what's in a doctors head and could use the experiences of thousands of doctors to do a diagnosis.

The problem with that though would be the legal ramifications of something like a misdiagnosis, or delayed diagnosis.

Say a person was told their cough was just allergies, but actually turned to be something like lung cancer. Could a doctor then off put the blame on to a program they used to help with making their original diagnosis?

With that put you, the developer, at risk?

Also, what would make your program so much better than a doctor simply searching up their suspicions on something like the internet?

What specifically could Machine Learning do that that couldn't?

A phd and prior history of starting billion dollar companies.

Machine learning algorithms could absorb more information in a few days than humans could in their whole entire lives. That's the main advantage.

Legal ramifications wouldn't be a huge problem. Who ever is providing the diagnosis service would be held responsible if something that wasn't supposed to happen happened. Doctors make mistakes too don't forget that. I'm sure the legal problems can be worked out if everyone agreed that machines were much better than humans at diagnosing patients.

>Machine learning algorithms could absorb more information in a few days than humans could in their whole entire lives. That's the main advantage.
So do you want to have something a doctor can type symptoms into, and have return possible causes?

That would be a step in the right direction.

Maybe the algorithm could return the % chance that a certain disease or abornomality may be present

Of course the input doesn't have to be limited to symptoms Any tests such as blood tests could be inputted as well

I don't think they'd necessarily need help with identifying what could be the cause of abnormal levels of some compound/element/ionic compound in fluids, in addition to other symptoms.

Unless you're trying to provide a cheaper healthcare option to people (which could totally be a goal), I just don't see how this is any better than the system in place now.

However, going to see a specialist can be very expensive. I could see maybe a doctor drawing on this to help give someone a more informed diagnosis maybe? But even then, not sure how much people would trust an algorithm over a human specialist, at this point in time.

ML will be not much help. Real challenge is detection with low PPM levels. We already have sophisticated models/simulations of major and interconnected cellular pathways of pathological importance.

If you really want to explore the cutting edge - it’s neither computer science nor biology, interestingly. It’s going to come from physics - photonics. Look into nanobiophotonics and nanosensors.

Light is our weapon against all deseases !!!

>Dunning Krugers like OP think they know something experts in the field don't

classic, when you get into the real world you realize intelligence is a hinderance, not a help.

Another area of exploration would be to reprogram existing/natural nano machines such as microbiomes to detect target enzymes or proteins. This is the area of Synthetic Biology.

In fact, miniaturising spectroscopes (look at Raman Surface Spectroscopy) could be a good first step. Liquid Chromatography is already used in lab with great success. But those machines are large. Making it to Doctor’s chamber is a challenge.

Holy shit OP
You are going to invent IBM Watson.
Though if you need MD PhD on board I'm in for some shekels.

These guys are not investing in detection. Deseases are great business.

Ifu can get people to believe ur cancer scanning scam sure lmao, remember to exit scam tho or theyl find out