ICAI: The Labs – AI for Food in the NL

In February, ICAI: The Labs is focused on AI for Food in the Netherlands. The AI for Agro-food lab and the AI for Bioscience lab each present their work and discuss challenges and developments made in this field.

AI for Agro-food lab is a collaboration set up by Wageningen University & Research Center together with the three other universities of technology in the Netherlands, OnePlanet and several industry partners. The aim of the lab is to evolve a new generation of ecology-based sustainable agricultural production systems that are supported and enhanced by smart tailored technology.

AI for Bioscience Lab is a collaboration between Delft University of Technology and DSM. The lab focuses on improving production technologies and developing bio-based products using AI.

Program

12.00 (noon): Opening

12:05: Introduction of the AI for Agro-food lab by Congcong Sun (WUR)

12:15: Rekha Raja (WUR) will present and demonstrate: ‘agro-food robotics’.

12.25: Introduction of the AI for Bioscience lab by Marcel Reinders (TU Delft)

12:30: Chengyao Peng (TU Delft) will present ‘Toward microbiome-based precision feed for animal nutrition’.

12.45: Discussion what’s next in AI for Food in NL

13.00: End

Abstract technical talk AI for Bioscience lab

‘Toward microbiome-based precision feed for animal nutrition — Linking feed and animal phenotype through microbiome’

Modulating the animal gut microbiota through precision nutrition is a promising strategy to improve
animal phenotypes, including greenhouse gas emission, feed conversion rate, animal health and
product quality. However, the associations of the animal gut microbiome and various important
animal phenotypes remain unclear. For the past decade, metagenomic analysis has been widely
adopted to establish such associations. Nevertheless, this approach faces enormous challenges due
to the resulting compositional data and the complexity of the associations. Lately, machine learning
methods, including supervised learning such as regression and classification, have shown the
potential to tackle the challenges of the problem. Here we present a regression model that works on
metagenomic compositional data to delineate associations between rumen microbiome and
methane emission. The preliminary results show that the model can explain around 12% of the
methane emission variance in our cattle samples. From the resulting coefficients, potential target
microbes in microbiome-based precision nutrition are prioritized.