Machine Learning-assisted Research on CFS

[Tunguska]

@Tunguska
The thread on Unfolded Protein response is outdated, i have posted several times that findings (as hypotheses) have progressed since then and that the only thing remaining right now is to have Medical Researchers evaluate the hypotheses being generated by the Machine Learning Algorithms

It's just that nowadays I have trouble seeing how those hypothesis are even worth considering if they're not formulated based on ideas from understanding of the biochemical pathways.

You could get lucky, but relevance of pathway isn't based on occurrences in research, but on estimation of the biochemical contribution of the pathway in question, first from understanding their role and relative importance in general human health, and then factoring in the specific disease parameters. There's a ton of popular pathways that are total wild goose chases. Also I think including non-CFS/ME research into this sort of meta-analysis almost invalidates it alone, because even when you understand the biochemical pathways the other disease research can very easily throw you off because it's extremely complicated, so with meta-analysis I don't see how you can expect anything but garbage (no offense).

 

[mariovitali]

@Tunguska

No problem, could you elaborate some more on the "you could get lucky" part of your post?

Do you mean that the topics chosen in the network graph can be attributed to pure chance?

 

[Tunguska]

@Tunguska

No problem, could you elaborate some more on the "you could get lucky" part of your post?

Do you mean that the topics chosen in the network graph can be attributed to pure chance?

It's what Jonathan Edwards said. The research topic trends and popularity aren't very meaningful, at worse they're completely misleading. It's a bet on popularity, plus being relevant to a specific disease at the same time (this part I find especially alien because CFS/ME has opposite parameters to other diseases, and I wouldn't equate it to gulf war, PFS or accutane exactly). If you look at the limitations and utter failings of accepted medical knowledge, nevermind that specific to CFS/ME, together with the state of and motivations for medical research, let's just say I hope you didn't bet your house.

Granted I don't grasp the intricacies of NLP (I have some background in computer science but I sure didn't do a masters in NLP), but it's hard to imagine it makes a difference.

 

[mariovitali]

@Tunguska

To be honest i did not quite understand what @Jonathan Edwards meant by "trending topics". I was under the impression that he was referring to the use of Machine Learning to Medical Research as being a trend nowadays.

The other thing that i could think of is that i am using Topics already being used (=Trending) in CFS/PFS etc research trying to prove something? I really do not understand so i will not hypothesise this further. A clarification would be very helpful.

Would you also like to comment that @A.B got CFS after Accutane use? How about the other case of the 18-year old girl using Accutane?

Again, i am not suggesting that Accutane did this since i am not related to any Medical Profession. This is a task for Medical researchers, not me.

 

[Tunguska]

@Tunguska

To be honest i did not quite understand what @Jonathan Edwards meant by "trending topics". I was under the impression that he was referring to the use of Machine Learning to Medical Research as being a trend nowadays.

The other thing that i could think of is that i am using Topics already being used (=Trending) in CFS/PFS etc research trying to prove something? I really do not understand so i will not hypothesise this further. A clarification would be very helpful.

Would you also like to comment that @A.B got CFS after Accutane use? How about the other case of the 18-year old girl using Accutane?

Again, i am not suggesting that Accutane did this since i am not related to any Medical Profession. This is a task for Medical researchers, not me.

I'll just comment on the accutane: I also took it and it might have been the original trigger for my CFS/ME symptoms or downward spiral leading to them. For this reason I also read a lot about it and some related conditions.

I don't believe CFS/ME and PFS are that similar. The PFS might have two different manifestations, but mostly what it shares with CFS/ME are brain alterations (I'm very interested in this part). But outside of the brain, I don't think the manifestations e.g. skeletal are very comparable.

Accutane is in between. It's documented to both alter the microbiome and wreck the hippocampus. If you entertain the possibility that CFS/ME might originate in the gut, you could see how accutane might cause CFS/ME. On the other hand if it doesn't wreck your gut, the brain degeneration and hormonol changes would manifest more like PFS.

(I'm ignoring liver changes here because although it's probably part of all their pathologies, from what is documented it's not likely to be identifying and as such I doubt addressing it is a cure)

 

[adreno]

I don't believe CFS/ME and PFS are that similar.

Agreed. The symptoms don't look similar at all.

 

[mariovitali]

@adreno

I will disagree.



The following post shows a good list of features :

http://www.propeciahelp.com/forum/viewtopic.php?f=3&t=10387


Sexual
[X] Loss of Libido / Sex Drive
[X] Erectile Dysfunction
[ ] Complete Impotence
[X] Loss of Morning Erections
[X] Loss of Spontaneous Erections
[X] Loss of Nocturnal Erections
[X] Watery Ejaculate
[ ] Reduced Ejaculate
[ ] Inability to Ejaculate / Orgasm
[ ] Reduced Sperm Count / Motility

Mental
[X] Emotional Blunting / Emotionally Flat
[X] Difficulty Focusing / Concentrating
[ ] Confusion
[X] Memory Loss / Forgetfullness
[ ] Stumbling over Words / Losing Train of Thought
[X - on rare occasions] Slurring of Speech
[X] Lack of Motivation / Feeling Passive / Complacency
[ ] Extreme Anxiety / Panic Attacks
[X - yes, somewhat] Severe Depression / Melancholy
[ ] Suicidal Thoughts

Physical
[X] Penile Tissue Changes (narrowing, shrinkage, wrinkled)
[X] Penis curvature / rotation on axis
[X] Testicular Pain
[ ] Testicular Shrinkage / Loss of Fullness
[X] Genital numbness / sensitivity decrease
[ ] Weight Gain
[ ] Gynecomastia (male breasts)
[ ] Muscle Wastage
[ ] Muscle Weakness
[ ] Joint Pain
[ ] Dry / Dark Circles under eyes

Misc
[ ] Prostate pain
[X] Persistent Fatigue / Exhaustion
[ ] Stomach Pains / Digestion Problems
[ ] Constipation / "Poo Pellets"
[ ] Vision - Acuity Decrease / Blurriness
[ ] Increased hair loss
[X] Frequent urination
[ ] Lowered body temperature

[ ] Other (please explain)


Chronic Fatigue :

https://en.wikipedia.org/wiki/Chronic_fatigue_syndrome#Signs_and_symptoms



Unrefreshing sleep, Fatigue/Exhaustion ,constipation, Loss of memory concentration, Joint Pain, Depression, Brain Fog common in both Syndromes.

Low body temperature found across posts in PR, also exists in the list of PFS. I am sure more similarities can be found.

EDIT : And here is Erectile Dysfunction :

https://www.researchgate.net/public...me_A_nationwide_population-based_cohort_study

 

[Snow Leopard]

Question: If you were in my position, would you move forward with this requirement as set forth by the responsible person of the Research team under these circumstances?

Yes, I would move forward.

Otherwise there will be no progress.

But candidate gene SNP studies on their own are not valid, there has to be some direct (subsequent) evidence of disturbed biochemical pathways, with a strong association with disease/symptoms for the theory to be credible.

 

[Tunguska]

I sort of agree with the cat guy, I mean I'm coming off overly negative, but if you can prove me wrong, or determine a valid hypothesis despite the method, then by all means do it.

But I absolutely think that equating CFS/ME with PFS and mixing theories about them - most should see this from the list you posted - is going to keep people from taking you seriously.

 

[Maya24]

What did the machine learning algorithm say about CBT / GET ?

CBT + GET = Error 404. Page not found

 

[Maya24]

But on a more serious note. Im actually really excited about finding a solution with machine learning. I dont completely agree that frequency of subject signifies importance. It could very well be that the subjects which appear smaller might be the ones who deserve the spotlight.

 

[mariovitali]

@Maya24

I also do not agree that frequency of subject signifies importance. This is not what is happening here and Machine Learning does much more than that (in my opinion).

I cannot possibly expect from people having nothing to do with Machine Learning, Natural Language Processing and Network Analysis to understand what is happening and how these Algorithms suggest candidate topics.

I would expect that they will neither completely dismiss it or completely accept it.

I believe that things are quite simple : A number of Genes have been given. Researchers may look at these Genes (and associated Pathways) in any way they believe is relevant. They may then dismiss these Genes as being irrelevant (and thus the results presented here) or they may find that these Genes appear to be relevant and pursue this Research further.

@Tunguska

I never equated PFS and CFS or suggested anything similar to what you are writing.

 

[Tunguska]

I never equated PFS and CFS or suggested anything similar to what you are writing.

Ok?... I have to take your word on that, because everything you're writing, up to "CFS/PFS" suggests otherwise. I politely suggest making it clearer how you're delimiting these diseases; better sooner than later...

As far as the better minds looking at this can tell, PFS is mostly a CNS (hippocampal) disorder affecting hormonal balance and response, perhaps reciprocally with the hormones in circulation. CFS/ME is a body-wide blood-bourne factor, immune or otherwise, causing a "dauer"-like state involving failure to oxidise glucose through oxidative phosphorylation and resulting somehow in body-wide PEM. These don't appear to have any causal relation. The symptoms match this. The CNS similarities are extremely intriguing but they can just as easily be compared to Alzheimer's and other diseases.

 

[Maya24]

@mariovitali I realise it's more complicated than a keyword search. The learning curve for any newly diagnosed CFS sufferer can be quite daunting.
I trust that someone somewhere will find the answer.

 

[Jesse2233]

@mariovitali why not make three models? one with CFS / PFS and two with each on their own?

It might be interesting to compare the outputs each one arrives at, and it would let other researchers pick what they thought was best to follow up on

 

[Valentijn]

Great, Thanks !

Exactly. Instead of learning more about a core component of your own theory and treatment, criticisms of short-fallings in that area are flippantly dismissed. I don't think you're taking this nearly as seriously as you need to to form a coherent hypothesis, much less prove anything about it.

To tie in SNPs regarding causation and cure, you really need to learn how they work.

 

[mariovitali]

@Jesse2233

I believe that this would raise the complexity (among other things which i cannot discuss at present). In any case i would keep an open mind about it and perhaps i could try this in the near future. But i think that at this point it is imperative that an evaluation is being made by close cooperation with Medical Researchers.

@Tunguska

This is your interpretation i am afraid. Since 2015, i am discussing about a common basis that these syndromes may potentially have (hypothesis) and not for all or some of these syndromes being the same.

It also appears to me that you are confident on what really is behind CFS and PFS : If you know then please make a statement about it.

Again, i do not suggest that these syndromes are the same, i suggest that they have a common basis. I really cannot say anything more if you repeatedly fail to understand the difference.

Nevertheless, could you comment -especially since you also took Accutane!- on the same topics i pointed to @Valentijn in this post ? http://forums.phoenixrising.me/inde...sted-research-on-cfs.51283/page-2#post-852531

 

[mariovitali]

But candidate gene SNP studies on their own are not valid, there has to be some direct (subsequent) evidence of disturbed biochemical pathways, with a strong association with disease/symptoms for the theory to be credible.

I Agree (and liked your post for this reason).


However it is a start, after which the hypothesis can be pursued further. Do you agree on that?

 

[Snow Leopard]

However it is a start, after which the hypothesis can be pursued further. Do you agree on that?

Yes. The key is to propose simple testable hypotheses - if you are in contact with a local team who is capable of good quality research, then this is your best opportunity to have the hypotheses tested.

 

[mariovitali]

OK, This may be interesting perhaps.


First, a discussion about Lupus, Rituximab and CFS :


https://www.healthrising.org/forums/threads/rituximab-lupus-and-chronic-fatigue-syndrome.3889/

You may also find a lot of hits on PR about Lupus.



Then we have the following (Note that GAS6, TYRO3, AXL, MERTK are mentioned):

Vitamin K-dependent proteins GAS6 and Protein S and TAM receptors in patients of systemic lupus erythematosus: correlation with common genetic variants and disease activity

<SNIP>

Systemic lupus erythematosus (SLE) is a chronic autoimmune disease characterized by an impairment of the regulation of the immune system [1]. This is reflected in hyperactivity of lymphocytes, the production of pathogenic auto-antibodies, and the formation of immune complexes, which can lead to multi-organ damage. Although the exact aetiology is not known, one of the possible triggers of this autoimmune reaction is a deficit in the process of efferocytosis, the clearance of apoptotic cells [2]. Under normal circumstances, apoptotic cells are engulfed by macrophages and other phagocyting cells using specific mechanisms, different from those of phagocyting bacteria and other corpuscles [3]. This provides a clearance of possible auto-antigens in the early phase of cell death without inducing inflammation or the immune response. In SLE, however, the clearance of apoptotic cells by macrophages is impaired, which may allow apoptotic cells to serve as immunogens for the induction of autoreactive T and B cells and drive the production of auto-antibodies [4].


The reasons for the defective clearance of apoptotic cells in SLE are not clear. The past decade has provided significant evidence that the complement cascade, immunoglobulin (Ig) M or pentraxins as serum amyloid P-component and C-reactive protein, contribute to the clearance of apoptotic bodies. Macrophages recognize apoptotic cells through an array of surface receptors [3]. Among them, the Tyro3, Axl and MerTK (TAM) tyrosine kinases play an especially important role in the clearance of apoptotic cells by macrophages and dendritic cells [5]. Mice lacking the three TAM receptors rapidly develop lupus-like symptoms, being the MerTK receptor the most potent mediator in this instance [6, 7, 8]. The main ligands that bind to and activate the TAM family of receptors are growth arrest-specific 6 (GAS6) and protein S (ProS), which are also capable of binding to negatively charged residues exposed early in apoptosis on the membrane surface of the apoptotic cell [9].


Conclusions
Here we show the association of several parameters related to the activity of TAM receptors and their ligands in patients with SLE. As suggested by other reports, GAS6 and other components of the system could be good markers of the development of SLE. Further studies with larger populations are needed in order to validate this assumption, as well as to determine the possible genetic components that could be best linked to the disease.

Link : https://arthritis-research.biomedcentral.com/articles/10.1186/ar4199

I also see several posts on PR mentioning SLE, so any comments much appreciated!



PS : @JaimeS what do you think?