Predictive Analytics Models in Vet Med

  • 23 Jul 2018 2:46 PM
    Message # 6394227

    I didn't see any topics in here so thought I'd share a project I've been researching and hoping to implement.

    I'm working on implementing an out-of-the-box Sepsis Predictive Analytics computing model from a human EHR (Epic) which leverages 70-80 granular data points within the patient's chart to predict likelihood of having or developing sepsis. This will [hopefully] replace a current physician advisory that only looks to 4-5 granular data points in the patient chart with a model that purports to have significantly higher positive/negative predictive values.

    Simply out of curiosity, is anyone aware of veterinary PIMS software companies that are looking to leverage granular patient data to predict disease process or outcomes for early intervention? 

    I wonder if this type of cognitive computer modeling would get more traction in food animal production environments?

  • 24 Jul 2018 8:29 AM
    Reply # 6395371 on 6394227
    Kash Kuruppu
    Hi Eli,

    Capitol Diagnostic Informatics is a veterinary data firm in Canada ( that provides analytical services to practitioners and will be launching a suite of cloud-based software for veterinary analytics in the near future. 

    Hope this helps,

    Kash Kuruppu, DVM

  • 30 Jul 2018 8:29 AM
    Reply # 6404469 on 6394227

    I'd be extremely curious to know what specific clinical disease process they were focusing on first? Or, is this an on-demand service with no out-of-the-box modeling available. 

    When you are able, I think everyone would be interested in learning the nuts of bolts of how they extracted patient data to improve patient care. Most of the analytics I've run into, for Small Animal Clinics at least, have focused on business measures and less on clinical measures.

    Thanks for sharing!


  • 07 Aug 2018 1:17 PM
    Reply # 6420885 on 6394227
    Kash Kuruppu

    Hello Eli,

    I very much appreciate your interest. Your observation is correct in that most veterinary analytics providers have solutions based on business measures rather than clinical measures. There are three main reasons for this:

    1. Health data is not standardized and data is stored within independent databases that are difficult to integrate. 

    2. Health data is complex. There is unique and redundant information collected by various care providers and stored in a variety of formats.

    3. Health analytics require access to patient records while maintaining confidentiality and privacy of patients.

    One of CDI's approaches to address these concerns is by providing a self-service platform to swine practitioners. Users onboard existing swine health data that they would like to analyze using our pre-built, swine health-focused algorithms   without having to share data with a third party. This app allows swine practitioners to manage risk and minimize losses by early detection of disease processes.

    If you are interested in learning about similar services, Verinovum is a service that provides a data enrichment and interchange platform for human patient data. AgConnect mHealth is a veterinary focused tool for collecting and reporting syndromic data. 

    Please feel free to get in touch with me directly at if you have further questions. 

    Kash Kuruppu, DVM

Google Group            

Powered by Wild Apricot Membership Software