Re : Norway Patent for ME/CFS treatment using Oxalate

hmnr asg

Senior Member
Messages
179
Likes
379
@mariovitali
Are you looking at just the abstracts ?
Also what is the result of this hypothesis? Are you finding correlation ? And if yes how does that prove causation ? I mean just because you dumped a large corpus of text into an NLP algorithm and found some co-occurence it doesnt mean you found anything of value .
I'm sure I will find a lot of co-occurence between the terms "Micheal Jackson" and "propofol", but that doesn't mean propofol causes Micheal Jackson*.

* He died as a result of a propofol overdose

H
 
Last edited:

mariovitali

Senior Member
Messages
1,076
Likes
1,096
@mariovitali
Are you looking at just the abstracts ?
Also what is the result of this hypothesis? Are you finding correlation ? And if yes how does that prove causation ? I mean just because you dumped a large corpus of text into an NLP algorithm and found some co-occurence it doesnt mean you found anything of value .
I'm sure I will find a lot of co-occurence between the terms "Micheal Jackson" and "propofol", but that doesn't mean propofol causes Micheal Jackson*.

* He died as a result of a propofol overdose

H
Good question.


Your way of thinking makes sense but it’s a bit simplistic. There are already companies doing this and in a nutshell , we are finding both known and unknown relationships between Medical topics of interest.

Unfortunately I cannot go into the details but so far , this technology was able to identify key aspects of ME/CFS (Pyruvate Dehydrogenase, Bile Acid metabolism, Fatty Acid metabolism, possible Liver issues) before any other Researcher.

Given my complete lack of knowledge in Medical Science I would say that these are quite interesting events that deserve further evaluation.

The software looks at the abstracts only.
 

hmnr asg

Senior Member
Messages
179
Likes
379
Good question.


Your way of thinking makes sense but it’s a bit simplistic. There are already companies doing this and in a nutshell , we are finding both known and unknown relationships between Medical topics of interest.

Unfortunately I cannot go into the details but so far , this technology was able to identify key aspects of ME/CFS (Pyruvate Dehydrogenase, Bile Acid metabolism, Fatty Acid metabolism, possible Liver issues) before any other Researcher.

Given my complete lack of knowledge in Medical Science I would say that these are quite interesting events that deserve further evaluation.

The software looks at the abstracts only.
The only rebuttal to my criticism that you gave me was that my view is too simplistic.
Please send me ONE , just ONE , credible paper from any legitimate machine learning/NLP conference where the cause of, or the cure to, a disease was found purely by running an ML algorithm on pubmed.
Also your proof for the validity of your algorithm is just that it has had some outputs that have ended up being relevant to CFS according to latest research.

The root of my skepticism is that no ML algorithm can synthesis new knowledge out of nothingness. Can you run your algorithm on pubmed and find the cure for cancer ? The elixir of eternal youth ? Also if you build a knowledge graph of medicine CFS it stands alone as a singular node. What structure is there for an ML algorithm to discover?

H
 

mariovitali

Senior Member
Messages
1,076
Likes
1,096
@hmnr asg

Your way of thinking is simplistic because you do not know the technology. The example with Michael Jackson is an unfortunate one in the sense that NLP and the discovery of associations do much more than what you think you know.

In the example you provided, NLP will output that propofol is associated with deaths (given there are many entries to suggest that) and not that propofol causes Michael Jackson. Given the right implementation NLP and NLU (Natural Language Understanding) provides the context of the association. If you knew how this technology works you would know that.

I will post a more detailed answer to your scepticism soon.
 

hmnr asg

Senior Member
Messages
179
Likes
379
So far you are just saying I lack any understanding.
That may be true but please let the code/paper be the arbitrator of this disagreement.
I don't work in ML but I have a PhD in electrical engineering. I sometimes read ML papers and I think I get the idea most of the time. Also I'm ok good with Python ( I'm assuming you used Python like most NLP folk these days ?) Also I know my deep learning in case that is what you have used here.

So please don't talk to me like I'm a 6 year old. My Micheal Jackson example might have been too simplistic but I have no idea what you are doing and I just hope to come to some understanding.

H
 

mariovitali

Senior Member
Messages
1,076
Likes
1,096
@hmnr asg

Again, I find that you are overly confident to a subject you do not know very well.

Most “NLP Folks” do not use Python but specialised software for implementing NLP techniques in a proper manner.

If you would take a closer look to the snapshots I provided you would see that I probably use a Python package called scikit-learn which has nothing to do with Deep Learning. If you really had knowledge on this you would have known from the name of the algorithms.

Please do not criticise in such a way something you do not quite understand.
 

hmnr asg

Senior Member
Messages
179
Likes
379
I think we got off the wrong foot.
I sincerely apologise if I have seemed insulting or minimized your hard work in anyway . I am a CFS sufferer myself and I really appreciate any effort on the part of anyone who is trying to find a cure. So I very much appreciate your hard work as well. I don't work in NLP but I really like to understand! If I seem overconfident please forgive me. Before my CFS started i did tend to be arrogant and these days when my old tendencies resurface I just remind myself that I am merely a shell of my former self and even reading more than an hour causes me a crash. So please believe me when I say I'm not over confident ! A lot of meaning gets lost over text and so I apologize again and I thank you once more for your hard work.

And I know about scikit-learn, I also know its not used for deep learning. But again, my experience is second hand and if you did decide to give me some added information it would be great . And if not that's also fine. I'm sure you have better things to do than to argue with a guy who rarely leaves his bed :(

H
 

mariovitali

Senior Member
Messages
1,076
Likes
1,096
No problem,

If we say that the technology is a black box , we keep only where it points at. In this case it points to the Liver and to specific pathways which I forwarded to several Researchers.

I am trying for the past 3 years to have them look at this hypothesis and to test it under strictly controlled conditions.

I cannot possibly know if this technology is correct but there is evidence that it may well be, at least for a subset of ME/CFS patients. I do not find the fact that people after many years of problems got symptom free ( this includes my self) a random event

If you wish to know more about the implementation PM me, I will be happy to help you out but have in mind that the methodology used is Patent-pending so there is a limit to what I can disclose :)