Comparative Veterinary Informatics Workshop
Corralling Veterinarians and Ontologists to Improve Health Outcomes
Harmonizing human and veterinary clinical and medical research data has the potential to improve health outcomes across species. While many advancements have been made with human phenotyping and precision medicine, the wealth of veterinary data available about domestic and exotic animals is not yet standardized, and it has not been leveraged in the same way that human data has, limiting our ability to perform similar analyses on animals, or to translate our knowledge about animal diseases into knowledge about human health. Recently, a diverse group of experts in fields ranging from human health informatics to canine cancer, came together to share their experience and knowledge in these fields, and to build a roadmap for future collaborations. Members of the Monarch Initiative from the Center for Health Artificial Intelligence at the University of Colorado Anschutz, in collaboration with faculty from the Colorado State University College of Veterinary Medicine and Biomedical Sciences (CVMBS), organized a workshop on Comparative Veterinary Informatics, which counted with 40 participants in a combination of in-person and online attendance. The workshop was sponsored by an NIH BD2K conference grant “Forums for Integrative phenomics” and CSU One Health Institute. Workshop goals included building an engaged community dedicated to advancing veterinary informatics, reviewing the “nuts and bolts” of ontologies and semantic interoperability, and brainstorming ways to make veterinary electronic health records (EHR) more interoperable, discoverable, and reusable.
Data Cat-egorization with Ontologies and Knowledge Graphs
Veterinary data is collected in heterogeneous formats such as EHRs, biobanks, and other sources, making it difficult to integrate the disparate data for the purposes of comparison and analysis. To offer effective examples of knowledge representation and organization, the workshop began with an overview of ontologies and knowledge graphs. Ontologies are formal classifications of knowledge that assign permanent, unique identifiers to concepts in a given domain; these concepts are both textually and logically defined to provide semantics or meaning to terms. Ontologies allow for the hierarchical classification of the concepts they contain. They are cross referenced to related resources, and illustrate the significance of each concept to the field they represent by including definitions and, in some cases, even layperson synonyms. Ontologies, together with knowledge graphs, allow for integration of heterogeneous data into a unified, queryable structure. The semantic standards community is eager to standardize the available veterinary clinical and research data to make it more discoverable and reusable in the context of efforts that already exist (for example, using existing biomedical ontologies such as the Unified Phenotype Ontology (uPheno) (see presentation) and the Mondo Disease Ontology).
State of the Arf Veterinary Informatics Updates
During the course of the workshop, participants were invited to share details of their work and how it is helping to advance the goals of reaching the promise of precision medicine in animals and successfully leveraging outcomes of veterinary research to advance our knowledge of human health and disease as appropriate.
We began with updates on the development of the new Vertebrate Breed Ontology (VBO), an important step in the standardized classification of animal breeds. Including breed, phene, and disease ontologies in the Online Mendelian Inheritance in Animals (OMIA) repository will allow for cross-species analysis of health data in an open, global, standardized, and computable format. For example, developing canine breed, phene, and disease ontologies are important steps to operationalize the field of comparative oncology. Similarly, defining a feline phenotype ontology and developing genetic tools for precision medicine based on the 99 Lives Cat Genome Sequencing Initiative will improve feline health and aid in future study.
Among the overlapping interests of the bioinformatics and veterinary science communities are the establishment, maintenance, and utilization of biorepositories such as the Flint Animal Cancer Center Biorepository at CSU and the Cornell Veterinary Biobank. Biobanks require quality management systems to ensure high quality biological specimens and associated data for research and precision medicine applications.
Other areas of shared interest are the standardization of companion animal genotype/phenotype data for translational research and understanding how companion animals can be bio-sentinels for environmental exposures associated with cancer and other health risks. For example, elucidating the molecular mechanisms of inherited cardiomyopathies in companion animals may lead to improved prevention and treatment options in humans and animals. Darwin’s Ark Cancer Project is a multimodal study that utilizes owner surveys, EHRs, environmental sampling, and genomics to expand the sample sizes of canine cancer studies with the goal of improving outcomes for canine and human cancer patients. Because humans and pets share the same household environment, there is a lot of interest in the potential of companion animals as sentinels for relevant exposures. A recent workshop organized by the National Academies of Sciences, Engineering, and Medicine aimed to understand the effects of environmental exposures on aging and cancer susceptibility in companion animals and humans. An important outcome of that workshop was the need to focus on environmental justice and inclusion of people and pets of lower socioeconomic classes, where the effects of environmental pollution may be amplified.
Bringing together human and veterinary data for One Health research, a transdisciplinary approach to advancing human, animal and environmental well-being, is made easier by the adaptation of veterinary electronic health records (EHRs) to the Observational Medical Outcomes Partnership (OMOP) common data model (CDM) at Tufts Cummings Veterinary School (TCSVM), Colorado State University CVMBS (CSU), and UC Davis Veterinary School (UC Davis). The TRanslational ANimal Shared CoLAboraTive Observational Research (TRANSLATOR) project enables querying for human and animal health data across all sixteen CTSI One Health Alliance (COHA) institutions. An example of a potential outcome includes an understanding of household (human and companion animal) antimicrobial resistance patterns. The platform at TCSVM has also allowed for the creation of the Tufts Splenic Tumor Assessment Tool to provide clinicians with a probability of malignancy estimations in dogs with splenic masses.
Finally, a wealth of information is available in the Zoological Information Management System (ZIMS) database. Standardized data on the care and welfare, husbandry, medical care, and studbooks for greater than 22,000 species and 10,000,000 animals over 44 years is fundamental to curating knowledge in conservation and comparative medicine.
With so much active research among the participants, the conversation then turned to the challenges of data integration and the need for interoperability to accelerate innovations.
Workshop participants had the opportunity to make connections and brought together different perspectives on short and long-term needs for advancing the field. This workshop provided a unique venue to build collaborative transdisciplinary teams and discuss potential strategies for improving standards, tooling, and communications.
The specific goals of this community are to:
- Leverage what is known about animal disease to improve human health by harmonizing EHRs to allow for household-wide (human and companion animal) analyses. As well, combine animal health data across all COHA institutions to improve machine learning for veterinary medicine and create a pathway for applications in human medicine.
- Advance veterinary care by focusing on precision medicine concepts from human health. Science is capable of doing for animal health what we do for human health in terms of preventing, diagnosing and treating inherited diseases, particularly as genetic sequencing of companion animals becomes more widely available; we should strive to achieve the same success with applying the accomplishments of precision medicine in humans.
- Advance understanding of genomic and mechanistic under-pinnings of disease in humans and animals using data-driven approaches.
- Empower pet owners and guardians to participate in their pet’s care through a combination of phenotyping, improved access to genomic testing, and streamlined transfer of knowledge.
Comparative Veterinary Informatics Workshop
When: May 24–25, 2022
Location: University of Colorado, Anschutz Medical Campus, Aurora, CO, USA
Organizers: Nicole Vasilevsky, Anne Thessen, Monica Munoz-Torres, Sue VandeWoude, Jori Leszczynski, Melissa Haendel, Sarah Gehrke, Brandon White, Julie McMurry
Agenda: Available here
Slides and Recordings: Available here
Funding: This workshop was supported by an NIH BD2K conference grant “Forums for Integrative phenomics” to Melissa Haendel and Peter Robinson: U13CA221044, and a generous donation from the CSU One Health Institute.