The Ur-Quan Masters Home Page Welcome, Guest. Please login or register.
Did you miss your activation email?
February 24, 2018, 08:37:18 pm
Home Help Search Login Register
News: UQM development migrated from Subversion to Git

+  The Ur-Quan Masters Discussion Forum
|-+  The Ur-Quan Masters Re-Release
| |-+  Starbase Café (Moderators: Michael Martin, fossil, Lukipela)
| | |-+  Machine learning for infrared spectroscopy
« previous next »
Pages: [1] Print
Author Topic: Machine learning for infrared spectroscopy  (Read 135 times)
Zanthius
Enlightened
*****
Offline Offline

Posts: 590



View Profile
Machine learning for infrared spectroscopy
« on: February 02, 2018, 10:55:08 am »



I am thinking that if we develop good HPLC-IR systems, then we can analyze our blood for vitamin definiciencies much more cheaply. Maybe we can even have such machines in our homes.  

I actually find it quite insane that most hospitals are using HPLC-MS systems, even though it probobly would be sufficient for them with much cheaper HPLC-IR systems.
« Last Edit: February 02, 2018, 12:15:05 pm by Zanthius » Logged
Death 999
Global Moderator
Enlightened
*****
Offline Offline

Gender: Male
Posts: 3616


We did. You did. Yes we can. No.


View Profile
Re: Machine learning for infrared spectroscopy
« Reply #1 on: February 02, 2018, 04:29:22 pm »

That's a thought. I've definitely looked at spectra and thought, "I have no idea what this means."

I don't know if there's enough good information in the spectrum to get that out, though.
Logged
Zanthius
Enlightened
*****
Offline Offline

Posts: 590



View Profile
Re: Machine learning for infrared spectroscopy
« Reply #2 on: February 02, 2018, 06:03:40 pm »

I don't know if there's enough good information in the spectrum to get that out, though.

The famous chemist Robert Burns Woodward that won the nobel prize in chemistry in 1965, is supposed to have been able to read molecular structures from infrared spectra. However, even if we might be able to prove that it is impossible to predict molecular structures from IR-spectra with 100% certainity, it doesn't mean it is impossible to make highly reliable predictions. So a machine learning algorithm might for example say that it believes it to have a specific molecular structure with a 95% probability.

If however the machine learning algorithm only searches for mathces to specific molecules (for example vitamins), then it can probably do it with a much higher level of certainity. So for routine checks at hospitals it shouldn't be a problem.

There are databases with IR-spectra available on the Internet (https://www.researchgate.net/post/Does_anyone_know_a_good_and_free_online_database_of_IR_spectra), so anybody with an Internet connection can start making a machine learning algorithm for this right away.
Logged
Scalare
*Many bubbles*
***
Offline Offline

Posts: 243



View Profile
Re: Machine learning for infrared spectroscopy
« Reply #3 on: February 05, 2018, 02:58:29 pm »

Before we all get our hopes up that your idea can actually fix things:
What is your expertise in this subject matter?
Logged
Zanthius
Enlightened
*****
Offline Offline

Posts: 590



View Profile
Re: Machine learning for infrared spectroscopy
« Reply #4 on: February 05, 2018, 03:46:47 pm »

Before we all get our hopes up that your idea can actually fix things:

You shouldn't, unless you have a basic understanding of what I said.

What is your expertise in this subject matter?

This seems like an appeal to authority, which is a cognitive bias: https://en.wikipedia.org/wiki/Argument_from_authority

Quote
One of the great commandments of science is, "Mistrust arguments from authority." ... Too many such arguments have proved too painfully wrong. Authorities must prove their contentions like everybody else.

And here:

Experts vs. Dart-Throwing Chimps



https://investingcaffeine.com/2012/07/08/experts-vs-dart-throwing-chimps/
« Last Edit: February 05, 2018, 04:04:20 pm by Zanthius » Logged
Scalare
*Many bubbles*
***
Offline Offline

Posts: 243



View Profile
Re: Machine learning for infrared spectroscopy
« Reply #5 on: February 05, 2018, 05:03:30 pm »

Before we all get our hopes up that your idea can actually fix things:

You shouldn't, unless you have a basic understanding of what I said.

I am actually a well versed programmer with experience in machine learning. But not the medical/biological experience.

Quote

What is your expertise in this subject matter?

This seems like an appeal to authority, which is a cognitive bias: https://en.wikipedia.org/wiki/Argument_from_authority
So that means that we should trust you without you having authority?
Logged
Zanthius
Enlightened
*****
Offline Offline

Posts: 590



View Profile
Re: Machine learning for infrared spectroscopy
« Reply #6 on: February 05, 2018, 05:15:40 pm »

I am actually a well versed programmer with experience in machine learning. But not the medical/biological experience.

Well, I probably have more experience with chemistry than with programming.

So that means that we should trust you without you having authority?

No. You shouldn't trust me no matter what.
« Last Edit: February 05, 2018, 05:53:09 pm by Zanthius » Logged
Pages: [1] Print 
« previous next »
Jump to:  


Login with username, password and session length

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines Valid XHTML 1.0! Valid CSS!