Hacker News Reader

Comment by trash3 | Hacker News Reader

Is there a need to account for circadian rhythms in sampling, do you ask your customers to swab at noon or before bed, or do you just include this info in model training/prediction?

View on HN

At the moment, we don't account for circadian rhythms. As we explore some of the insights to the oral microbiome, that's definitely one of the avenues we'd love to dig into. We do ask customers to fill out a relatively detailed health survey, so that we can explore associations between the oral microbiome in health and disease.

> We do ask customers to fill out a relatively detailed health survey, so that we can explore associations between the oral microbiome in health and disease.

So, you're learning from customers these associations, and after a significant number of customers you could offer statistical associations to future customers?

A little bit of both actually - as our user base grows we create a network effect that will spur new statistical associations. We can apply those new findings to new and existing users by updating their report accordingly.

You can imagine that even if you took the test once and had (for the sake of simplicity) 1 insight being reported back, you could revisit your profile in 2 years when we have tested 10,000 users and may have 4 new insights that have been added to your report as a result of the new discoveries we've made!

TLDR; there are associations reported now, and new ones that will be reported as we grow our user base and data

David put it nicely (and more succinctly) in another reply: "In short, there is a network effect with our platform. Each sample we sequence contributes to a database of novel bacteria, viruses, and fungi that we can use to discover new microbes." I would just add that discoveries also apply to new associations that we can include in future reports.

Reply on news.ycombinator.com