In this article we continue to describe the collar measurement algorithms.
In veterinary science, there is no clear method suitable for determining ideal indicators for a particular individual. Even the calculation of the ideal weight of an animal is done “by eye” using the BCS method. Therefore, we are “forced” to determine which group a particular dog belongs to.
In this way, we take into account the following factors, such as:
height,
weight,
gender,
age,
activity level.
The next step is to do the same, but only with activities - that is, those that the dog does. Because it is clear that the pulse of a healthy dog, regardless of age, gender and weight, will be lower during sleep than during active training. Therefore, we need to understand what exactly the dog is doing when we measure its pulse, temperature or something else. There are many activity options, but we were able to identify several key categories into which different types of activities can be grouped. For example, sleeping or walking.
In the next step, we correlate all the measured indicators with dog categories and types of activity. In reality, it looks like this: “For category A1, the normal heart rate in the sleep state is from X to Y.” Hundreds and thousands of lines of categorization.
The next step is to formulate a diagnostic hypothesis. We are deliberately talking about a hypothesis. Only a doctor can make a diagnosis. We just assume that there are signs characteristic of a particular disease. For the most common diseases, we have created the so-called “diagnostic profiles” where the symptoms are “decomposed” in the format we need. For example, what symptoms could indicate possible signs of tachycardia (increased heart rate) in a dog? Obviously - increased heart rate. However, increased heart rate during exercise is the norm. Yes, of course. And an increased heart rate in puppies up to 9-12 months of age (depending on breed and size) is also normal. Therefore, we create a rather complex diagnostic profile that allows us to take into account many points.
The simplest thing is then that the system looks at the actual indicators of the animal, classifies the animal into one category or another and applies a diagnostic profile taking into account activity. For example, an animal of category A1 has an increased resting heart rate and SIMULTANEOUSLY an increased heart rate in a sleeping state OR SIMULTANEOUSLY a possibly increased temperature.
This is where the additional complexity of our algorithms lies - we look at all the data that we have SIMULTANEOUSLY. This allows us to increase the accuracy of the diagnosis.