Julie Rehmeyer's 'Through the Shadowlands'
Writer Never Give Up talks about Julie Rehmeyer's new book "Through the Shadowlands: A Science Writer's Odyssey into an Illness Science Doesn't Understand" and shares an interview with Julie ...
Discuss the article on the Forums.

Machine Learning-assisted Research on CFS

Discussion in 'General ME/CFS Discussion' started by mariovitali, May 8, 2017.

  1. mariovitali

    mariovitali Senior Member

    Messages:
    850
    Likes:
    716
    Dear All,


    Please find below the first post of the series where Machine Learning, Natural Language Processing and Network Analysis is used to identify Medical Topics relevant to CFS, Firbomyalgia, Post-Finasteride Syndrome and more.


    [​IMG]


    Network Analysis and several Machine Learning algorithms point to Vitamin K - related Genes as highly relevant (Hypothesis) to syndromes such as CFS, PFS (and more).


    [​IMG]




    Some of these genes are : MERTK, GAS6 and VKORC1


    Full text :


    http://algogenomics.blogspot.com/2017/05/machine-learning-nlp-and-network.html
     
    Sancar, Maya24, gabriella17 and 7 others like this.
  2. RogerBlack

    RogerBlack Senior Member

    Messages:
    891
    Likes:
    2,890
    [​IMG]

    The problem is hard, and you don't fix it with simple approaches that look at the literature without that literature being reliable.
    You can't find data that isn't there. There is a lot of research into the biology of CFS/ME - but it's all in poorly understood areas of biology. We do not have a complete understanding of the mitochondria, and what turning on/off various genes does, immune behaviour, small molecules that may poison stuff, ...
     
    Last edited: May 8, 2017
    GreyOwl, Learner1, Sancar and 8 others like this.
  3. mariovitali

    mariovitali Senior Member

    Messages:
    850
    Likes:
    716
  4. Valentijn

    Valentijn Senior Member

    Messages:
    14,281
    Likes:
    45,822
    If you mention dozens of things (as done in that post) or hundreds (as done in that thread), odds are pretty good that you'll randomly hit on something that will be mentioned again in ME research.

    I really don't see an explanation for their relevance, or why you think those particular concepts are particularly meaningful. I'm also confused by your hypotheses, since Hypothesis #1 isn't an hypothesis, and there's been no apparent attempt to test Hypothesis 2, or even understand the science behind SNPs.
     
    skwag, Skippa and MEMum like this.
  5. RogerBlack

    RogerBlack Senior Member

    Messages:
    891
    Likes:
    2,890
  6. mariovitali

    mariovitali Senior Member

    Messages:
    850
    Likes:
    716
    @RogerBlack


    No problem. Actually the ways that these algorithms work *may* also identify indirect relationships between Medical Topics, in other words they may find previously unknown knowledge.

    I tried to be neither optimistic or pessimistic in this post. The post simply lists a number of Genes and then if any researchers find this list interesting then they pursue this further.
     
    TiredSam likes this.
  7. trishrhymes

    trishrhymes Senior Member

    Messages:
    2,153
    Likes:
    17,872
    I think we need all the scientific tools we can get to crack ME. If machine learning analysis of scientific papers brings up some new directions for researchers to study, surely that's a good thing. My knowledge of methodology is not sufficient to see whether this particular technique will ultimately help or not. It seems worth trying.

    I wonder why some of the information has been hidden on the diagram. Can anyone explain? Are the developers of this model afraid someone will 'steal' their ideas? Surely all evidence should be shared openly in good science.

    By the way, if you want to contact an ME body in the UK, I'd try Invest in ME Research or the ME Association rather than Action for ME.
     
  8. Skippa

    Skippa Anti-BS

    Messages:
    841
    Likes:
    2,960
    Agree with @trishrhymes chuck everything we can on the fire, something will catch!
     
  9. Valentijn

    Valentijn Senior Member

    Messages:
    14,281
    Likes:
    45,822
    Preferably things which are at least flammable. The treatment recommendations in @mariovitali's previous thread were based on a lack of comprehension of how SNPs work, and the presumption that all of them are relevant to the function of the gene they are on.

    I think there needs to be a basic understanding of the subject matter before there's any chance of progress being made. And I doubt that all of the diseases listed have a common cause or treatment, based on studies that show a fair bit of distinction between them.
     
  10. mariovitali

    mariovitali Senior Member

    Messages:
    850
    Likes:
    716
    Just got a message from Mrs Gersuk from OpenMedicineFoundation. She said that she will share these results with the team.

    This is great...

    I also got an answer from the PFS Foundation, saying they will also consider the results and get back to me.

    @trishrhymes Thank you for the info, i will contact them. The information that is missing from the diagram will be forwarded to a number of Researchers that have nothing to do with any of these Syndromes.

    As soon as there is an acceptance on the validity of these results, i will share them (if there is interest of course) to any CFS/PFS Researchers. The node in red located to the left of LXR (Liver X Receptor) is the one that is more central than any other node (you may also notice it is more red than any other Node).

    There is no reason to be sharing information that is most likely irrelevant to CFS/PFS but touches a more general subject.
     
  11. A.B.

    A.B. Senior Member

    Messages:
    3,751
    Likes:
    23,193
    I like that someone is using this approach to look for clues, but why hide some information?
     
    skwag likes this.
  12. BurnA

    BurnA Senior Member

    Messages:
    2,086
    Likes:
    9,863
    What did the machine learning algorithm say about CBT / GET ?
     
    arewenearlythereyet and Alvin2 like this.
  13. mariovitali

    mariovitali Senior Member

    Messages:
    850
    Likes:
    716
    @A.B.

    I do understand and i would probably think the same way if i were you. All i can say at the moment is that any findings related to the most central node may be disclosed only after proper evaluation by certified Medical Practitioners / Researchers.

    Recall that this Research is not only about CFS but several more syndromes.


    @BurnA

    Nothing as these are not among the Topics that the Algorithms considers ( currently 592 Medical Topics)
     
    A.B. likes this.
  14. A.B.

    A.B. Senior Member

    Messages:
    3,751
    Likes:
    23,193
    There are doctors and researchers on PR too. Just saying ;)

    Hiding this information just limits collaboration.
     
    JaimeS, skwag and mariovitali like this.
  15. mariovitali

    mariovitali Senior Member

    Messages:
    850
    Likes:
    716

    If these Researchers are specialists on Liver function, i'd be more than happy to show them. I do not think that Doctors would be of much help to be honest.
     
  16. BurnA

    BurnA Senior Member

    Messages:
    2,086
    Likes:
    9,863
    Thanks but how is it decided what topics it does consider.
     
  17. aquariusgirl

    aquariusgirl Senior Member

    Messages:
    1,426
    Likes:
    440
    I took a quick look ..,lines up for me.

    After years screwing around w/methylation,I am now leaning towards the idea that it is my sulfation that is the major problem & that is switching off methionine synthase.

    Oxolates...vitamin K ....bile synthesis .....yup. That's all explained by oxolates.

    Lots of gallbladder issues , ultrasounds, low secondary bile acids on tests ...,

    I posted where glucose, glycated proteins & lipid perixides converts to oxolates in the presence of free copper & iron ...,

    THe Alzheimer's researchers are looking at free copper..,

    https://www.ncbi.nlm.nih.gov/m/pubmed/15781823/
     
    anni66 likes this.
  18. mariovitali

    mariovitali Senior Member

    Messages:
    850
    Likes:
    716
    @BurnA

    Any medical topic can be used. As an example, consider that after looking at some Research papers i believe that P450scc (CYP11A1) could be relevant.

    The next step is to collect all PubMed entries on CYP11A1/P450scc and then have this Topic scored for its relevance algorithmically.
     
    Last edited: May 8, 2017
  19. adreno

    adreno PR activist

    Messages:
    4,843
    Likes:
    11,028
    Vitamin K is involved with sphingolipid synthesis which were found to be low in ME/CFS by Naviaux.
     
    Tunguska, A.B. and mariovitali like this.
  20. wastwater

    wastwater Senior Member

    Messages:
    959
    Likes:
    653
    uk
    Several types of bacteria exist as part of the normal flora of the human gut, including e coli,this bacterium has many beneficial functions such as the production of vitamin k and b vitamins
    That was a quote I saw in a medical paper not sure if it's right
     
    Last edited: May 10, 2017
    arewenearlythereyet likes this.

See more popular forum discussions.

Share This Page