Poster Topical Area: Nutritional Epidemiology

Location: Hall D

Poster Board Number: 820

P20-146 - Creation of an online interface facilitating personalized nutrition interventions based on genomics, metabolomics, proteomics, and microbiome data using datamining

Monday, Jun 11
8:00 AM – 3:00 PM

Objectives: Disclosing genetic and metabolomic information has been shown to increase compliance to dietary interventions (DIs) decreasing the risk of noncommunicable chronic disease. Personalized nutrition (PN) based on genomics and metabolomics is gaining increased attention. The objective of this project was to build an online interface to create PN actions based on genomic, metabolomic, proteomic and microbiome measures for patients and care providers using most recent information available.

Methods: We curated a database associating genetic SNPs (n>100,000), blood measures of metabolites (n>150) and proteins (n>200), and microbiome phylum data as well as adverse health conditions (n>100) with DIs and dietary patterns, by employing datamining software (Ovid for Medline) on human studies alone. Each DI was linked to specific food categories and food items, respectively, using the Canadian Nutrient File, to create specific PN actions. We developed machine learning algorithms, including an evidence-based confidence score, to rank DIs, and food categories within and across multiple measures.

Results: To date, genetic polymorphisms, metabolites, and microbiome phyla have been included in the database as being associated with DIs. MTHFR 677CT and glucose are the polymorphism and metabolite, respectively, associated with the largest number of DIs (n: 4 and 9, respectively) and the highest confidence score. Bacteroidetes-to-Firmicutes ratio was the microbiome measure with the highest confidence score and associated with n=1 DI. Inflammatory bowel disease was the adverse health condition associated with the largest number of DIs (n=7). Employing machine learning in a custom online interface ranked DIs according to their associations with multiple abnormal measures of metabolites, microbiome phyla and genetic polymorphisms based on patients' own data; this process enables efficient determination of individualized DIs with the highest potential to benefit. The tracking app further facilitated translation of DIs into PN actions that can be easily recorded by patients using a smart-phone based app to monitor and capture adherence longitudinally.

Conclusions: We created an online interface facilitating PN interventions and will determine compliance and efficacy in a longitudinal study.

Funding Source: This project was funded by Molecular You Corp, which is a personalized health company.

CoAuthors: Mohammad Anwar – Molecular You; Tya Hariharan – Molecular You; Nadya Calderon, MSc – School of Interactive Arts and Technology; Simon Fraser University; Grace Goh – Molecular You; Ana Marcu – Molecular You; David Wishart – University of Alberta; Solveig Johannessen – Molecular You; Robert Fraser – Molecular You

Theresa H. Schroder

Molecular You
Vancouver, British Columbia, Canada