The inaugural vHive event was held on the 11th December 2023 at the University of Surrey Vet School. The seminar consisted of presentations from local and international speakers, as well as an interactive discussion session, introduction to vHive and the launch of our new vHive Incubator. With permissions from speakers, we are able to share the following presentations:
Genomic Inbreeding with imputed SNP: Debate, lessons learned and future perspectives
Dr Christos Dadousis- Research Fellow in Health and bioinformatics, University of Surrey
Leveraging New methods in artificial intelligence for veterinary medicine
Professor Kevin Wells- DataHub director and professor of AI in Human and veterinary healthcare, University of surrey
Q1: Can this technology (syringomyelia) be applied via a smart phone?
A: Yes, there is some technology using deep learning that would allow this. It is our hope that the next piece of funding will allow us to investigate this further.
Q2: What’s the correlation between the MRI/CT image and the external image?
A: There is a very strong correlation. The potential factor which makes it not a perfect correlation is the fact that dogs have fur/hair on the surface and different amounts of muscle on top which can be a confounding factor. But there is an extremely strong correlation of the distortion between the MRI, shape on the CT and the external images.
Q3: Why did you train your vet chatbot on social media?
A: It was trained on a whole variety of things but essentially it was created using product engineering which is a way of making the model play a particular role.
Q4: Are there AI tools being applied in vet medicine looking at pathology diagnosis e.g. looking at slides etc.
A: There is some of that, but the bulk of the work on digital pathology is in the human area. This is a particularly exciting prospect- There are great opportunities under the ‘one health’ banner, to be able to transfer human health care over to the animal health domain.
Q5: Is that a simple lift and shift from human into veterinary medicine? Are there any barriers or obstacles to overcome in making that translation?
A: Different tissue types present in different ways so there are some challenges, but we’ve seen so far that it is a very viable thing to assume.
Translational diagnostics in companion animals with emphasis on microRNAs
Dr Susie Armstrong- Senior lecturer in veterinary clinical research, University of Surrey
Q1: How could we transform any of these tests into ‘pen side’ tests?
A: There are a couple that I’m working with that are handheld monitors so would just be a case of loading up the sample then getting a reading within half an hour to an hour? The really exciting thing about this is that they can be multiplexed- We’ve got another handheld device that will test any protein there is and they can be multiplexed. I think we are going towards signatures, so it’s not going to be one MIR its going to be a signature of MIRs and we can then use the AI to improve the sensitivity and specificity.
Q2: What is the depth of the micro RNA data- in terms of the number of entities that you’ve measured/quantified?
A: We’ll load up a sample and have 10 million reads then we hit across the sequence and probe the different gene sequences. As dogs aren’t completely sequenced, we take the dog sequence, then the human and mouse sequences and build our picture from there.
Q3: If you did a bit of sieving of the data, you could start to compare different diseases?
A: Absolutely. I’m really cognisant of this- we use the same methodology throughout so that all of my studies are comparable. We use the same kits- I extract the RNA with a procedure/protocol developed in Edinburgh, then they go to the same sequencing lab to use the same sequencing kits to build our library to make sure they are comparable. There is no doubt that this is a problem within the field- the comparability between studies.
Understanding High-pathogenicity avian influenza outbreaks using data and genomics- AI for AI
Dr Samantha Lycett- Senior lecturer/ group leader in pathogen phylodynamics, roslin Institute
Q1: Is there any hope for rare species (Canadian condor) that have been adversely affected by new strains?
A: I think part of the problem is that they’re catching AI through predation of other species. There is a lot of risk of spill over. Some countries are thinking of vaccination (e.g. domestic ducks in France). People are wondering whether these domestic species can be vaccinated. There isn’t any licenced vaccine in the UK at the moment, however there is a vaccine of suitable strains under development to be used in places like China which could be useful. With the rare species, you generally know where they are and how many, so there may be the potential to vaccinate them.
Q2: Avian Influenza- is it a fast or slow mutating virus?
A: Its quite fast, typically you get 5 mutations per thousand sites per year.
The use of aI and statistics in optimising the use of livestock big data
Professor guilherme Rosa- Professor in Precision Livestock Production and Breeding, University of Wisconsin-Madison
Q1: There is a huge issue with the animals dying in transit. Is there an opportunity to acclimatise the pigs to the transport so they’re used to the truck?
A: I’m not sure if they do anything like this with the pigs. In the second part of my talk where I talk about the pigs moving within the compound, these are very short distances but still have the same stressful experience. So it may be more stressful to acclimatise them.
Q2: We have big issues with both human and animal data software and sources/databases. You’re using massive data sets. How do you clean up your data into a useable format?
A: We get the data from different sources, but we export everything ourselves. The farm data is from the software at the farm, then we have the weather data that we download from the web, all in different forms. Quite often we have image and sound to analyse but we use the same software to analyse from different sources. It is a big chunk of the project to pull the data together.
Q3: The real value of big data is to embed that culture right down to even the smallest farmers. Do you think there is a way we can encourage data collection and sharing such that we can realise the true value?
A: This is very important but difficult. It is working within the industry (dairy or beef) to then work with producers. It is difficult to do this one by one, maybe via associations? We need to find a way to share data but it’s complicated.
If you have any questions for our speakers about their presentations, please email us using the email below