Piavita provides a digital, remote monitoring solution with the Piavet System for the management of a wide variety of cases. And one of our key vital signs is body temperature.
The Piavet Measuring Device uses sensors to calculate a horse’s body temperature (among other values) and sends that data to the Piavet Base Station and then to the Piavet Cloud Applications where machine-learning is applied to ultimately display precise vital sign information on the Piavet Platform.
As algorithms need data to learn and become more accurate, it’s important we gather certain data (and a lot of it!) to improve the accuracy of our vital sign measurements. In October, we set out to do just that for body temperature by launching the Body Temperature Improvement Project.
Fever in horses
Fever is not a disease but merely a symptom of an existing illness that helps to fight pathogens. However, if it rises too high, it can have very negative effects on a horse’s health, as it attacks the protein molecules of the body.
Fever can be triggered by the immune system or an infection. Horses with fever show reduced appetite and thirst, may separate themselves from the herd, or appear lethargic. If a horse owner or caretaker recognizes these signs, it’s important that action is taken quickly to reduce the fever and investigate the cause. However, this is one reason why gathering fever data in horses can be challenging.
Why focus on temperature?
Horses with fever are often classified as intensive care patients. And these patients need as much rest as possible. But this can be challenging to ensure, as there is often a lot going on in a clinic that a patient may find disruptive, including regular but necessary vital sign checks. But as we know from our veterinarians, temperature is one of the most important parameters to assess a patient’s health status. With non-invasive, long-term vital sign monitoring, the Piavet System enables vets to provide the patient with rest and extend control intervals.
Since we measure temperature non-invasively (not rectally), the temperature measurement with the Piavet System can differ from that of a thermometer. However, if the values are within the normal range, our measurement is already very accurate, with only +/-0.1 °C deviation. But due to small amounts of “abnormal” temperature data currently feeding the algorithm, it’s difficult to predict fever and hypothermia. With these cases, our algorithm can currently display a trend. Meaning, if the horse’s body temperature rises due to fever, the displayed temperature on the Piavet Platform will also rise. However, the deviation from the actual body temperature will increase.
What will be improved?
Our machine learning algorithm currently uses sensor data from heart rate, changes in heat radiation from the body, skin temperature, and environmental temperature to calculate body temperature. The temperature is then displayed on the platform as an average of the last 20-second measurement window. By working together with a Swiss company that has produced temperature sensors for human healthcare, we have found the perfect partner for this project. The basic idea will not change, but the machine learning approach will now be based on a neural network, which is a more evolved artificial intelligence system. The new model will be able to make more accurate predictions.
In order for our partners to adapt the machine learning model, they need a certain amount of fever data with reference values from rectal measurements of different horses. This is where our Piavet Community (both customers and industry partners) play a big role. Some of our Piavet users have been diligently collecting and submitting data since October, and we are thrilled to be partnered with Purdue University on the project as well! But as an algorithm is only as good as the quality and amount of the data that feeds it, we are still in need of more data!
How to participate
Collecting fever data is easy. If you are a Piavet user, follow the steps below and look forward to enjoying better patient monitoring, even for your most critical patients.
- 1. Enter all required patient information on the Piavet Platform when creating the new patient (click here for platform tutorials).
- 2. Secure the Piavet Belt, soak the coat thoroughly with alcohol, and insert the Piavet Measuring Device into the placeholder.
- 3. Start a measurement and select the case Temperature Improvement Project. Note: The ideal measurement is at least three hours long.
- 4. Measure the patient’s rectal temperature regularly and often as possible.
- 5. Record rectal temperature values in a note directly in the measurement at the time they are taken. Note: This will give us the reference data we need and is a vital step!
That’s it! So, if you have a feverish patient, don’t hesitate to put on the Piavet System and collect the data. And while you’re doing that, enjoy digitally and remotely monitoring your patient for other signs of distress.
For an in-depth tutorial on how to collect fever data with the Piavet System, see our video below.
While this project will result in a more precise temperature measurement for our Piavet users, there is another benefit! By submitting fever data, you will also automatically collect PiaPoints, which can be used toward big discounts on your annual user fees. That’s what we call a win-win!
Looking to the future
By participating in data collection using the Piavet System, you are helping the equine veterinary industry as a whole. Piavita is at the forefront of digitalization in veterinary health monitoring, and our loyal customers are right there with us. Their active support—from using the Piavet System in their daily practice to participating in development projects like this one—is paving the way for the future of technology in veterinary medicine.