The recommendations are based on studies or trusted sources.  Studies do not always produce the same effect — diet, disease and a dozen other factors may change the conclusion for a study. To illustrate this, let us look at berberine, listed at this page: http://microbiomeprescription.com/Library/GutModifier/?name=berberine

Looking at the results we see some consistency across multiple studies:

• Bacteroides Increases Source: . Sixty out of the 134 OTUs we…
• Bacteroides Increases Source: Berberine, a major pharmacolog…
• Bacteroides Increases Source: Phylum/Genus mean ± SD mean ± …

We also see some slight disagreement:

• Blautia Increases Source: . Sixty out of the 134 OTUs we…
• Blautia Increases Source: Berberine, a major pharmacolog…
• Blautia Increases Source: DataPunk.Net…
• Blautia Decreases Source: Phylum/Genus mean ± SD mean ± …

And others which are a toss up…

• Lactobacillus Increases Source: Compared to their decreases in…
• Lactobacillus Decreases Source: Phylum/Genus mean ± SD mean ± …

The AI processing of the recommendations attempts to balance these factors (I hope to get the next generation coded this weekend). The recommendations are not guarantees of being effective, just items with better odds.

Items with multiple names

Different studies may describe what may be the same item in different ways. For example:

• resveratrol
• grape seed
• grape seed extract
• grapes
• red wine
• red wine polyphenol
• red wine polyphenols

Or

• green tea
• tea

And for other cases — spelling difference

• vsl#3
• vsl-3

or

• salvia officinalis
• salvia officinalis (sage)

or

• gluten free
• gluten-free
• gluten-free diet

This can result in some odd recommendations, with one being on Avoid and one on Take. As the data gets better cleaned up, we should see less of these.

Bottom Line

“When in doubt, leave it out”