Veterinarians get data-savvy to map livestock health
Innovative data-driven tool will accelerate collection of crucial evidence
Good data on livestock disease and mortality is essential for making informed investments in animal health, and to improve productivity and incomes in low and middle-income countries. But decision-makers face enormous data challenges when it comes to grasping and making sense of the existing body of literature and evidence, which is scattered across multiple databases or even unpublished. To address this, the University of Edinburgh is bringing together world-class veterinary knowledge with cutting-edge informatics and industry expertise, to develop an automated tool to extract and summarize evidence from academic publications.
Since 2017, researchers from Supporting Evidence Based Interventions (SEBI), based at the University of Edinburgh’s Royal (Dick) School of Veterinary Studies, have been working to determine which livestock diseases are prevalent in a handful of sub-Saharan African countries, and the percentage of animal deaths due to disease. Their approach to compiling the best available evidence includes a rigorous “systematic mapping review” process that assembles the existing body of literature. This labour-intensive task involves scouring different databases for published studies or grey literature, deciding which studies to include or exclude, and then painstakingly extracting country-specific data on the prevalence of different diseases and associated mortality rates for different livestock types. It can take over three months for a researcher to extract the relevant data and build a report on relevant livestock disease for one country.
To accelerate and improve this process, SEBI have partnered with researchers at The Bayes Centre and SAS, the leader in analytics. Together, they will innovate the way that information and evidence is extracted from research papers relating to livestock data in sub-Saharan Africa, by developing an automated text analytics tool that has potential to reduce a three-month process down to a matter of hours. This is the first project by the School of Veterinary Studies to be funded by the Edinburgh and South East Scotland City Region Deal’s Data-Driven Innovation initiative, which supports innovations for economic growth, social change, and public services. The activity is co-funded by SEBI, which is supported by the Bill and Melinda Gates Foundation.
“The need for evidence-based decision making is more pressing than ever, and decision-makers need comprehensive, robust and rapid evidence,” said Dr Karen Smyth, Deputy Director of the SEBI Programme. “This tool has the potential to improve the quality and efficiency of our work. We estimate that we could extract and analyse data from over 60 countries in the time it would normally take to do one country,” she said.
While systematic reviews in other disciplines are already using automation tools, different types of systematic reviews have different requirements, especially when it comes to handling large datasets. The animal health literature in low and middle-income countries is particularly disparate, and constantly growing, and calls for a novel approach.
“The fusion between informatics and the ever expanding data available to the agritech and veterinary communities is an exciting and timely development,” said Bruce Whitelaw, Data-Driven Innovation Agritech Lead and Deputy Director of the Roslin Institute. “It moves us towards the goal of accelerating the flow of information towards evidence-based decision making,” he said.
To start with, SEBI researchers and their donors will work with informatics experts to ensure the solution can be scaled across multiple projects. The veterinary scientists will then work with text analytics experts to translate their requirements into an automated algorithmic solution. The final product will be a user-friendly software application that rapidly searches and screens databases for relevant literature, extracts the specific data of interest, and produces reports on livestock data.
“This collaboration will fulfil an important need,” explained Dr. Jasmina Lazić, Chief Data Technologist at The Bayes Centre. “Not only will we accelerate the review process but we will be able to create a live, updatable evidence database,” she said.
The collaborators also hope the tool could eventually be used to train the wider academic research community to extract information and evidence from research papers.
“There is huge potential to extend this solution to other research groups across the University and could even be used in a commercial setting,” said Dr. Lazić.
“Ultimately this tool will be a resource for academics, governments, and anyone who needs up-to-date information on the prevalence of livestock diseases in a country,” said Dr. Smyth. “This is critical evidence that can be used to improve the well-being of smallholder livestock keepers, particularly in Africa,” she said.