This is a subset of content available on GitHub, focusing on samples collected from the same company at the same time. This also includes code to reproduce plots and statistics.
Again, I took no action based upon any of these reports.
Intuitively, I thought something about the variability for the same sample for the Viome collections seemed high. I think the plot above might help illustrate this.
You can also see more about the underlying details below:
|
Company:
Signature / Score |
Mean |
SD |
|
Viome:
Active Microbial Diversity (Percentile) |
6.5 |
7.8 |
|
Viome:
Digestive Efficacy |
64.0 |
9.9 |
|
Viome:
Gas Production |
38.0 |
11.3 |
|
Viome:
Gut Lining Health |
67.5 |
0.7 |
|
Viome:
Gut Microbiome Health |
45.0 |
5.7 |
|
Viome:
Inflammatory Activity |
38.5 |
3.5 |
|
Viome:
Metabolic Fitness |
28.5 |
0.7 |
|
Viome:
Microbiome-Induced Stress |
43.5 |
10.6 |
|
Viome:
Protein Fermentation |
35.0 |
9.9 |
|
thryve:
Gut Diversity Score |
93.5 |
0.7 |
|
thryve:
Gut Wellness Score |
81.0 |
0.0 |
To be fair, some Viome signatures / scores may be relatively robust. Also, in general, every individual claim needs to be investigated independently (and you not should draw conclusions about one analysis based upon what you saw for a different analysis, whether that is good or bad). However, I think there is evidence to back up my conclusion that you should not follow all recommendations, and treat the overall set of scores/signatures as hypotheses that may or may not be absolutely true for specific sample.
On the original GitHub page, I think there is some additional evidence (looking at measurements for all 3 companies) that increased variability in similar samples makes it harder to detect a noticeable phenotypic difference in the 4th sample. However, to be fair, I think that is less rigorous than the comparisons that I am making above.
Posts
No comments:
Post a Comment