Machine Learning-assisted Research on ME/CFS

@Ravn

I believe that HRV is a very important predictor so bought a heart monitor strap and connected it to my phone. I got my first reading (see attached) which is considered a low reading. I also stared a drug which was proposed through Machine Learning . I know all the signs that my body gives and so i will know within a few days. I expect also that the HRV will increase over the next period of time.

In any case this is really exciting, i will post updates.

@Trish @Andy do you think that HRV monitoring deserves a different thread?

 

@wigglethemouse this will take some time but sounds interesting. Where did you find an entry for methyl glucopyranoside ? Is it in the CSV file from the supplementary material?

 

Yes, in my opinion a separate thread on your HRV would be sensible if it's not going to connect directly with your machine learning work.

 

@mariovitali Yes the Excel file published with the paper here.
www.mdpi.com/2218-1989/8/4/90/s1

Here is how I analysed the data. I crudely sorted the data in the Excel file by difference in median values between patients and controls to see what was interesting. That's how I ended up with similar top metabolites to those in the Volcano plot analysis in the paper.

Histamine was near the top and I plotted it here. Histamine is one of the mediators released by mast cells. In this small data set it looked interesting to explore further.
https://twitter.com/user/status/1071253664138846208


I crudely removed controls with histamine >210K and removed patients with histamine <210K from the spreadsheet to see what other metabolites jumped out when I resorted that table by difference in median (tweet shows 200K, for the life of me I can't remember whether I used 200K or 210K). PFOS screamed at me when I did that. Strangely there is no mention of PFOS in the paper.
https://twitter.com/user/status/1071279132682514432


This is PFOS plotted with the original data (no removing any data)
https://twitter.com/user/status/1091024605328793601


I read the EPA and news articles surrounding PFOS and there seems to be quite a bit of sweeping under the carpet and many folks say the "safe" levels are set way way too high - even a White House report. Many studies have been done on PFOS/PFOA and it's known to affect the thyroid. I never looked to see if there is a known effect with the liver.....

Anyway, back to
- taurodeoxycholate
- perfluorooctanesulfonic acid (PFOS)
- methyl glucopyranoside (alpha + beta)

These appeared at the top of my second table when I crudely manipulated the data for histamine. They had the biggest jump from my original ranking by median difference in the first table to the rank in the second table

Code:

Orig  New   Rank    Metabolite
Rank  Rank  Change
6     1     5       histamine
2     2     0       pimeloylcarnitine/3-methyladipoylcarnitine (C7-DC)
3     3     0       tauroursodeoxycholate
47    4    43       taurodeoxycholate
8     5     3       piperine
44    6    38       perfluorooctanesulfonic acid (PFOS)
34    7    27       methyl glucopyranoside (alpha + beta)
5     8    -3       3-hydroxybutyrylcarnitine (1)
16    9     7       heme
7    10    -3       3-hydroxybutyrate (BHBA)
14   11     3       taurocholate

This is the median/average values for the original data in my first table for methyl glucopyranoside

Code:

        Control Patient Difference
Median  231897  404242  1.743194177
Average 441114  485311  1.100192804

And this is it when I manipulated the data for the second table

Code:

        Control   Patient  Difference
Median   178085   519806   2.918861779
Average  365486   524010   1.433734473

I know it's very crude analysis, but I'm not a statistician. I could be way wrong in all this. Either way, thankful to Dr. Hanson and team for making the raw data available to play with.

Sorry if I rambled on too much in your thread.

 

So I guess my question is, if a large subset of patients are suspected to have mast cell degranulation issues, that could be indicated by the higher histamine values in some patients, can we use that knowledge to find other metabolites that may correlate, and what bodily systems are involved. From the top 10 shown in the table above bile acids and enviromental toxins could show a relationship to the liver. That's where I thought it would be interesting to get your take @mariovitali .

 

@wigglethemouse

Thank you, this is how we should be working...i will feed this data through some multivariate analysis methods and see what i come up with. There should have been more cases to be honest but i will give it a try.

 

Australia's Neil McGregor looked at this dataset, I think, and it was part of his presentation at the recent Emerge conference. Link to where he talks about PFOS as well as other things.

Code:

https://youtu.be/UvGMbbyGtvg?t=2076

copy and paste to go to the right time point in the video.

 

I must say that this puts a big smile on my face :

 

Thanks @Andy. You've got a good memory! He also mentions 4-hydroxychlorothalonil a fungacide. I hadn't seen that in the data.

What I did see in the "histamine" data and couldn't figure out was 4-hydroxycoumarin another another fungal metabolite
https://en.wikipedia.org/wiki/4-Hydroxycoumarin

I hadn't dug deeper on 4-hydroxycoumarin before as the patient/control sample size is relatively small. Food for thought.

Code:

Original spreadsheet
        Control   Patient  Difference
Median  95676     75147    0.785426857

Histamine manipulated spreadsheet
        Control   Patient  Difference
Median  95676     194982   2.037935323

 

I just sent an email to Janet Dafoe. I hope they consider this.


Apparently, Parenteral nutrition causes Choline Deficiency. If cholinergic signaling is indeed problematic in ME/CFS patients this can have serious consequences.



https://www.sciencedirect.com/science/article/pii/0270913995901434


Network Analysis, 2017 :

 

@wigglethemouse

I did try using specific ML techniques but i get incosistent results due to the small amount of data related to the number of features. I will look more on mast cells and get back to you.



Here is one interesting connection of tinnitus -a frequent symptom of PwME- with Vagal Nerve stimulation :

 

@mariovitali , sorry if this is something you have tried and reported on already, but can you, and have you, used your software to search for papers in other research areas that report on mitochondrial functioning being inhibited due to a factor carried by the blood? I have a loose understanding on what is possible so don't worry about telling me it's a silly suggestion.

 

There are no silly suggestions @Andy


I believe you are referring on the "something in the blood" mentioned by Ron Davis. I did try to see what i can find but this is as if i am trying to work with one hand only and allow me to explain :

This software framework should be used in Tandem with experts that know what they are doing. It is the experts that should be guiding the process and not someone such as myself who has no background in Medicine, whatsoever.

No one was listening since 2015 @Andy. If acetylcholine is indeed central to ME/CFS pathology and this was identified since 2015 then...i will stop here. Allow me to say once more that i will work with everyone, starting tomorrow, who is able to evaluate the findings of this tool.

It's sitting there, waiting. I do not know how soon i will be able to give these findings to be put to use. If anyone can help, please PM me.

 

I think there's a study of tVNS underway at Vanderbilt.

 

I made some really good progress today. It's all coming together when it comes to acetylcholine although it's very possible that this involves a subgroup of patients. We shall see.

I think the "liver disease" shown over and over in the algorithmic runs is there because without a proper liver function there is not much acetylcholine going around. The same when there is choline deficiency :




According to the following paper, there is a hepatic branch of the vagus nerve which is responsible to release acetylcholine

In other words, could a liver injury is responsible for dysregulation of acetylcholine production? There may be serious consequences down the line for sure, one involving the cholinergic anti-inflammatory pathway (CAP). From Wikipedia :

My HRV steadily rising, feeling much better , less bloated. I think i know why my health declined after i started dieting to shed some extra pounds. When lipolysis takes place, epinephrine kicks in and i probably do not want that (because it triggers the sympathetic nervous system !).

Regarding lipolysis. Observe "epinephrine" below :




Illustration of the activation of lipolysis in an adipocyte. induced by high epinephrine and low insulin levels in the blood. Epinephrine binds to beta-adrenergic receptors on the cell membrane of the adipocyte, which causes cAMP to be generated inside the cell.


I made more connections, will post in the next couple of days

 

@mariovitali . Are you aware that Systrom is trialling pyridostigmine? He's looking at cardio effect, but is there any way of looking at liver effect?

 

Just posting this here too :

I hypothesise that Glutamate along with Liver disease may be the missing pieces of the puzzle *if* a subset of ME patients does not respond to an acetylcholinesterase inhibitor assuming that such a medication is found to be ameliorating ME symptoms to some patients.

I also hypothesise that Glutamate is not beneficial for the symptoms of PwME

 

Something of interest, Patients on Phoenix Rising used Cayenne and had some nasty side effects :

https://forums.phoenixrising.me/threads/spicy-foods.54052/

and

https://forums.phoenixrising.me/threads/cayenne-pepper-disaster-black-out.46227/

Is it possible that Cayenne -which contains capsaicin - disrupts cholinergic signaling ? Non-human studies :

https://link.springer.com/article/10.1007/BF00504379

May be something to look for.

 

@wigglethemouse

I found the following connection for MCAS which i believe you are interested in :

The work of Robert Phair becomes -potentially- increasingly important for other reasons as well (based on my research).


https://www.researchgate.net/public..._activation_through_Aryl_hydrocarbon_receptor

 

I believed that i lost this one but i am glad i was wrong : Over the past days i am seeing a very significant improvement in word-finding and memory. The difference is very noticeable because i am able to recall compounds that i read very recently, something which i simply couldn't do.

I am gathering all the papers related to Kynurenine metabolism and will post here. I will also give a hypothesis on why i had a severe decline in my health and how all of this may be connected. If i had to "put my money" on current research, i would choose Robert Phair and Petter Brodin from Karolinska.

One example : Kynurenine metabolism and Liver fibrosis (in rats) :

https://www.nature.com/articles/s41598-017-11759-z