FightAMR: Novel global One Health surveillance approach to fight AMR using Artificial Intelligence and big data mining
Lead Research Organisation:
University of Nottingham
Department Name: School of Veterinary Medicine and Sci
Abstract
Understanding the risk and direction of antimicrobial resistance (AMR) spread through food-borne routes, and developing of interventions to limit the spread of AMR within and between humans, animals, environment and food is a significant challenge, requiring a 360-degree investigation of a complex, interconnected system of humans, animals, environment on one hand and geographical, societal and climate-related variables on the other.
This project will develop a monitoring system using AI and advanced tech to detect AMR spread in the interconnected human-animal-environment-food system ('One Health').
First, we will analyse the heterogeneous corpus of historical AMR-related public data. This will improve our understanding of what data (monitorable biomarkers) should be collected to identify the conditions leading to a higher risk of AMR spread. This knowledge will be used to guide a large-scale multi-country sampling collection campaign of a large amount of heterogeneous and interconnected data from farms, wet markets, food, environment. Data will include results of microbiological analysis, whole-genome sequencing, metagenomics, phenotyping, documentation of on-farm management practices, and environmental sensor data (temperature, humidity, etc). An innovative AI-powered data mining pipeline will be used to unravel previously unknown correlations between observable animal, human, environment, food variables and a core set of resistome, microbiome, and microbial genomics variables, highlighting new routes for surveillance deployable in low-to-high-income countries.
This project will develop a monitoring system using AI and advanced tech to detect AMR spread in the interconnected human-animal-environment-food system ('One Health').
First, we will analyse the heterogeneous corpus of historical AMR-related public data. This will improve our understanding of what data (monitorable biomarkers) should be collected to identify the conditions leading to a higher risk of AMR spread. This knowledge will be used to guide a large-scale multi-country sampling collection campaign of a large amount of heterogeneous and interconnected data from farms, wet markets, food, environment. Data will include results of microbiological analysis, whole-genome sequencing, metagenomics, phenotyping, documentation of on-farm management practices, and environmental sensor data (temperature, humidity, etc). An innovative AI-powered data mining pipeline will be used to unravel previously unknown correlations between observable animal, human, environment, food variables and a core set of resistome, microbiome, and microbial genomics variables, highlighting new routes for surveillance deployable in low-to-high-income countries.
Technical Summary
This project aims to develop new AI-powered surveillance solutions to identify increased risk of AMR emergence and direction of spread through the food-borne route based on the 'One Health' concept, and capable to detect the appearance of known and novel AMR traits. The solutions will be suitable for deployment in low-to-high income countries.
While there have been many studies on AMR in livestock, environment, and human, they often focus on a specific sector and/or rely on a specific analysis (antimicrobial usage or whole genome sequencing) but not necessarily on integrated data analysis. Our goal is to strengthen big data approaches to integrate surveillance across human, animal, and environment with the food chain to assist interventions to prevent AMR caused by resistant enteric bacterial pathogens.
We plan to devise a real-time monitoring method capable of pinpointing geographical locations and routes which -at any point in time- may be at higher risk of developing AMR. This will be achieved by a triangulated approach: i) Understand the conditions leading to higher risk of AMR, by developing a cloud-solution embedding auto-adaptive learning to integrate and mine public data from different sources (metagenomics, phenotypes, satellite, etc) at different scales (region/setting/country, etc) and subjects (humans, animals etc); ii) Perform an AI-guided experimental sampling collection campaign of unprecedented scale and coverage; iii) Identify monitorable biomarkers indicating increased risk of AMR and direction of spread, embedded in deployable surveillance solutions.
While there have been many studies on AMR in livestock, environment, and human, they often focus on a specific sector and/or rely on a specific analysis (antimicrobial usage or whole genome sequencing) but not necessarily on integrated data analysis. Our goal is to strengthen big data approaches to integrate surveillance across human, animal, and environment with the food chain to assist interventions to prevent AMR caused by resistant enteric bacterial pathogens.
We plan to devise a real-time monitoring method capable of pinpointing geographical locations and routes which -at any point in time- may be at higher risk of developing AMR. This will be achieved by a triangulated approach: i) Understand the conditions leading to higher risk of AMR, by developing a cloud-solution embedding auto-adaptive learning to integrate and mine public data from different sources (metagenomics, phenotypes, satellite, etc) at different scales (region/setting/country, etc) and subjects (humans, animals etc); ii) Perform an AI-guided experimental sampling collection campaign of unprecedented scale and coverage; iii) Identify monitorable biomarkers indicating increased risk of AMR and direction of spread, embedded in deployable surveillance solutions.
