An avian patient’s diagnostic results just came back. 100 toxic heterophils have been found. Can the veterinarian conclude that disease is present?
The question may seem simple enough that you immediately thought there must be some trickery involved. And, you’d be right as answering the question requires more context, and honestly, more data to make any clinical determination.
However, traditionally the presence of these toxic heterophils would be indicative of disease based on long-used reference intervals. But what if these reference intervals are incomplete to begin with? What if the presence of 100 toxic heterophils were actually normal for this bird based on its age, location, or even season the sample was taken?
Every journey requires a good map, but right now veterinarians are setting out on health expeditions without all the data or reference intervals to successfully complete the road to better animal health. They’re operating on potentially outdated conceptual maps, incomplete reference intervals, that act more as ideas rather than solid guides.
Compare the first map (top) of Great Britain based on Ptolemy’s Geography from 1475 versus Nicolaes Visscher II and John Overton’s 1685 rendition (bottom). Besides the fact that Scotland is placed to the right in the first, the second offers infinite more detail about towns and regions. Working with outdated reference intervals versus those based on accurate and modern data is very much the same; having a better map promises a smoother patient journey.
One principal way the veterinary community can meet this challenge is by creating reference intervals not based on class or order but based on species, and Moichor is well poised to meet this growing need.
What’s missing from the map now, or what have reference intervals traditionally left out?
Principally, the lack of species-specific reference intervals in animal clinical pathology is a major piece that is largely absent from what could be better maps to improved animal health.
So how is Moichor contributing to this map, and what are we doing differently now compared to several months ago when we released 36 new intervals?
Two key data initiatives have been set into play to grow our data set, and these include a greater effort in extending Moichor services beyond exotics to also include mammals and, consequently, providing novel and data-informed species-specific and even breed-specific reference intervals as a key product offering for practitioners and their associated clinics and institutions.
We’re map makers, cartographers of animal health, and we want to provide veterinarians with the tools they need to improve the patient journey. However, we’d be remiss in leaving our own journey of how Moichor came to this renewed reference interval commitment. We explain here how we came to this point, where we are, and where we’re going.
In the last three decades we've seen veterinary medicine move from mainly farm and animal herd health to companion animal medicine. People are also keeping a wider variety of species, and according to the Global Animal Health Association, these animals are living from 11.4% to 230% longer depending on species and location.
That shift has not necessarily translated into better reference intervals as veterinarians and practices don't generally have the time to create them. The time just isn’t there to carry out the research, so veterinarians decide to share data among species and stick with pathology providers who do the same.
Separately, as human medicine improves, veterinary medicine is expected to grow alongside it, and this can be daunting as we see unique therapies emerging for individual human patients while over 7,000 different species of birds who each have their own unique qualities are essentially treated the same when it comes to blood and chemistry work.
Firstly, our artificial intelligence allows us to not only perform automated cell counts but conduct automated cell morphology assessments as well. This means Moichor is able to identify and classify otherwise easily-missed variables with efficiency.
Secondly, in our young company’s time, we’ve amassed tens of thousands of tests, and among those tests are a large number of species-specific wellness samples from healthy patients. This means our team does not actively have to locate ideal animals for modern reference intervals — the data comes directly to us as part of our work with veterinarians.
Finally, we have a diverse team specializing in chemistry, clinical pathology, veterinary medicine, laboratory science, data science, and mathematics, all of which allows us to coalesce data together to create not just better trained artificial intelligence but also a better laboratory and, most pertinent to this discussion, better veterinary reference intervals.
One of the biggest steps Moichor has taken in building an accurate and modern reference interval library is to identify whether a test is part of a wellness sample or of a sick animal patient. Doing so allows us to strengthen inclusion and exclusion criteria and then conduct re-review from our pathologists.
In addition, if the team identifies healthy samples that could be smaller pieces of the larger species-specific reference interval puzzle, then Moichor veterinarians will reach out to the patients’ providers to let them know that their patients are part of the reference interval project. From there, veterinarians who are interested in contributing to the project answer surveys, share medical notes, go deeper into patient history, and jointly cover the nitty-gritty details that either preclude or qualify each patient.
Our team is key in creating these modern reference intervals, but our clients that are willing to help us really make this data powerful.
Consider this bigger example of toxic heterophils in birds listed at the beginning of this piece. Our team has been studying the same question posted above: is the presence of mild heterophil toxicity or band cells in birds a sign of disease?
We’ve been working to test the current general consensus around this question which is that any presence of said toxicity or cell immaturity is a sign of disease. How have we done this using AI? When we run our CBCs through the Moichor AI, this allows us to look at a massive volume of data. The data is then filtered through our inclusion and exclusion criteria using human intelligence from veterinarians and pathologists on our team.
After compiling data from our automated CBC and morphology assessments and pairing this with human assessments, what we are starting to find is that the presence of mild heterophil toxicity and/or cell immaturity may not always be indicative of disease. As we continue this investigation and confirm our initial findings, Moichor should be responsible for updating avian reference intervals to include a certain threshold for naturally existing toxicity as it appears in healthy species.
Reference intervals are maps to interpreting CBC and chemistry data. If the guide is more robust in reflecting the specific population that you're trying to assess as a clinician, then your assessment is going to be more appropriate and hopefully help with avoiding over or under interpretation. Ultimately, better maps lead to better patient care, and you no longer have to imagine the time and money saved as Moichor develops and publishes the most accurate species-specific reference intervals in veterinary medicine.