Why you MD knows almost nothing about the Microbiome

I view Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) — and many other conditions– as being either a pure microbiome dysfunction, or a microbiome dysfunction that contributes significantly to many of the symptoms. If you walk into your typical medical office and ask questions about this you will likely be met with a variety of responses (depending on their psychology) that are not helpful.

A reader asked “Why is this?” The easiest way to understand the issue is to look at the number of studies on pubmed about the microbiome

https://pubmed.ncbi.nlm.nih.gov/?term=microbiome&filter=years.2003-2021&sort=date&timeline=expanded

What does this translate to? Unless your MD finished his studies in the last 5 years, there is very little chance that there was any significant coverage of the microbiome in courses. Yes, they took microbiology — the study of bacteria; but the microbiome is different. How so?

Microbiology can be compared to understanding how a person‘s psychology changes as they ages from a child, to an adult, to an elderly. The microbiome is understand sociology of people in a society. The microbiome is the understanding of interactions and interplays. It is more complex. Far more complex.

This is why MDs and naturopaths tend to look for the single bacteria or virus responsible. That is precisely what they had training in, the comfort principle of medicine.

A second factor is simple, specialization. “Ah, the microbiome belongs to the human gastrointestinal system, so the patient should be referred to a gastroenterologist!” Wait! they treat very specific diseases only and not items like a diabetic’s microbiome. Tossing patients to specialists is a common practise — the problem is that there are rarely specialists in this area. It’s my least favorite game: PPP – “Patient Ping Pong”.

Old techniques versus Best techniques

Often old techniques could be rephrased as ‘current accepted best practises’. In my Uni days, I has several professors tell me “Do not go into Engineering or Medicine, you are brilliant — but you are also creative and innovative, that will end your career in those areas”. I have an additional characteristics, I am a high functioning ASD person. Why is that important? It means that I tend to ignore social pressure and conformity; instead, I march to my own drum beat along whatever path looks interesting (and have little anxiety about stepping off existing paths). Back to the topic….

What is the best technique for dealing with the microbiome? It is simple, use various types of artificial intelligence and machine learning. The problem with this for MDs is that machines are giving them advice that they are incapable of understanding the why. It is not that the MD is dumb, it is because the problem is very very complex.

One type of machine learning is called “Random Forest“. I have used them professionally when I worked for Amazon. Microbiome studies started using this in 2010 and the number of studies are exploding yearly — why, because it works!

https://pubmed.ncbi.nlm.nih.gov/?term=microbiome%20random%20forest&sort=date&timeline=expanded


The problem is that this is a new discipline called data science, an specialized application of advanced statistics. Yes, MDs and medical researchers often did a basic statistics class at Uni. The problem is that they have simplified what they were taught and incorrectly applied it.

A typical mistake that I have seen is reporting something like “the control group average as 30 and the treated group average was 50 with a less than 5% chance of being random”. Klaxon sounding!!!!

  • They looked at 100 different bacteria in their study, there is a good chance that 5 of them will have a 5% chance of being significant at random!
  • Often they will report on averages and appear to assume that the data is a well behaved normal or gaussian distribution.
    • On occasion, I have looked up the distribution of some of these “discovery” bacteria. The average was not at the 50%ile or median (expected with every normal curve) but at the 87%ile. The mode (most common value) was at the 10%ile.

In short, they are deficient in appropriate skills in handling numbers that arise with the microbiome.

Bottom Line

We need someone to fund serious state-of-the-art research into the microbiome and then evangelize the results into the medical community. This is a hot topic with many many microbiome testing firms being launched by venture capitalists. They see that this has the potential of being financially rewarding.

My own contribution is making a free site that uses data science and artificial intelligence on 16s samples. https://microbiomeprescription.com/ I have seen a few 16s firms, adapt/borrow, features from my site to their clients — I am very fine with that. We are just at the start of the microbiome journey – unfortunately, many still wish that this was just a walk in the park, instead of having to cross the Alps!