Data science is vastly changing the way we do research. Veterinary research is, however, still lagging behind its human counterpart. Why is this the case and can we learn from human health data to close this gap?
Published by: BioVox on November 27, 2024 - link
Written by: Marjoke Le Roy
Interviewed: Volodimir Olexiouk, PhD, Director of Scientific Engagement & Steff Taelman, PhD, Bioinformatic Scientist.
The possibilities of data driven veterinary science
Imagine a world where we could predict disease outbreaks before they happen, combat antibiotic resistance, or preserve endangered species, all by harnessing the power of data.
Leveraging data science in veterinary medicine would greatly improve both animal and human healthcare by contributing to the One Health principle. This highlights the close interconnection between human, animal, and environmental health, emphasizing their undeniable influence on one another. According to Volodimir Olexiouk, PhD, Director of Scientific Engagement at BioLizard, a Ghent-based data science company with extensive expertise in animal health data, improving animal welfare is our responsibility and data science can make a difference: “As humans, we have the responsibility to ensure the well-being of animals, to be respectful with animal products and to reduce our ecological impact. Through data science, we can gain deeper insights into today’s challenges and anticipate future ones, harnessing technology to drive meaningful progress.” However, compartmentalisation and scarcity of animal data currently prevent veterinary data science from reaching its full potential.
How data science can improve veterinary medicine
Industry and academic collaborations focusing on animal data science could be one solution to make veterinary science more data driven. BioLizard is a company supporting life science companies and research institutions in extracting actionable insights from (big) data. In one of their projects, researchers collaborated with the faculty of veterinary medicine of Ghent University to investigate the impact of feed type and gut microbiome composition on the occurrence of stomach ulcers in pigs, a very common issue worldwide. Using advanced bioinformatics and biostatistics proved to be crucial in inferring causal relations from microbiome data. Furthermore, quality control and preprocessing of qPCR and sequencing data enabled the removal of contamination from other species.
The analysis generated actionable insights which confirmed, for instance, that microbiome diversity buffers the negative effects of the pathogen that causes the ulcers. Thus, researchers hypothesized that probiotic supplementation might help reduce the occurrence of stomach ulcers in pigs, improving their well-being.
The hurdles to overcome
To generate accurate causal inferences, you need a lot of data which is currently one of the biggest challenges in veterinary data science. “Veterinary data can be very compartmentalized and scarce,” explains Steff Taelman, PhD, Bioinformatic Scientist at BioLizard. “Within human medicine, you have huge data sets from large consortia of universities, this is not as common for animal data.”
This problem has two main causes. Firstly, data is scarce because animal health is not tracked and information is not always shared as for humans. Where healthcare records form a vast amount of information for human research, they are not as widely used in animals. Farmers or animal health-focused companies might record health data of their animals, but this data is kept proprietary, compared to human health records which are kept in a shared database. Furthermore, veterinary research is published less frequently compared to human medical research, resulting in fewer opportunities to share valuable data, results, and insights. Together with the fact that animal health journals often don’t require research data to be published alongside the results, this creates a hurdle to building large animal databases.
Secondly, there is currently no regulatory framework governing animal data, resulting in compartmentalization. “Increased regulation would open a dialogue between companies,” explains Taelman. “This would prevent the repetition of failed experiments, support consistent conclusions, and aid data centralization.”
To tackle this problem, the European Medicines Agency (EMA) and Heads of Medicines Agencies (HMA) are running a veterinary big data initiative from 2021 to 2027. The strategy aims to build a data ecosystem, increase system interoperability and re-usability of data, and ensure data quality to make regulation of veterinary medicines in Europe more data driven.
The benefits of lagging behind
The lack of regulation governing veterinary data surprisingly also comes with some advantages. Together with quick ethical approval (compared to human research), it provides veterinary data science with a lot of freedom. It accelerates the approval for AI or machine learning-based experimental design optimization and sometimes even makes experimental animal data more accessible. For example, when deep sequencing the microbiome of pigs, you could also analyze the DNA of the host animals themselves without breaking any GDPR regulations, since they don’t apply to animals. This way, you can leverage more valuable insights.
Having a human data science predecessor also allows the veterinary field to learn from its insights, developments, and downfalls. Data science is no different for animals and humans, meaning the tools are fully translatable. “These technologies can be adapted into the veterinary field, significantly accelerating research and advancing the maturity of their data science capabilities.” explains Olexiouk. Human data itself can also give veterinary science a head start. Following the one health principle, it is common practice to look at how pathogens of interest and related microorganisms react in other animals, including humans.
One paw at a time
It all sounds very promising, but for veterinary data science to reach its full potential, more centralized and publicly available data is needed. Luckily, Olexiouk is optimistic for the future: “We are exponentially creating more veterinary data including images, genomics and more. The industry is seeing the benefits thereof, so I would only assume this will grow further in the future.”
