Using Artificial Intelligence in Horse Training
This morning my RSS feed alerted me to an article appearing on New Scientist titled "Wearable device for racehorses could help prevent fatal injuries"
After setting the scene in regards to the death rates of racehorse at racetracks (which made me think it was a purposeful press release to frame the product) the article went on to mention a new(wish) wearable device for horse training and monitoring called Equimètre and outline its claims as a product that will be useful in the Thoroughbred industry for training and health assessment.
The space for Equine health monitoring is fairly crowded at the moment so it is hard to see how successful something like Equimètre is going to be. The competition is stiff -
Well known health monitor Polar have been offering equine heart rate monitors for decades and many trainers and exercise physiologists use their products. The limitation being is that they only offer heart rate data and it is only for the time that the horse has a girth strap on them.
e-Trakka has been around for a decade or more and use a saddle pad to encapsulate a heart rate monitor and GPS that not only gives speed but also accelerometer data that allows for the measurement of stride length during training. The downside to it is that it is a saddle pad, which some trainers think is a little bulky.
EquinITy, a U.K based operation has a very similar process to it as e-Trakka but instead of a saddle pad, packs everything in a girth strap to be used with a saddle. It has the same heart rate, speed, and stride length measurement as e-Trakka and also allows for real time analysis.
GMax is another U.K based team that have a product similar to EquinITy although it must be said that it looks like that company has pivoted more into GPS based race monitoring with its partnership with Total Performance Data.
ClockitEq was an Australian based company that was bought out by KER in the US recently. They offer a cheaper version of Polar and e-Trakka with heart rate and time (sectionals) but no stride length.
SeeHorse is a Canadian based company that has built a wearable device that sits on the halter of a horse and allows the heart rate, temperature and respiratory rate to be measured continuously. Variations in these three variables can not only give a trainer a better idea of the response to training but also alert the trainer to variations that might be pointing to sickness or injury.
So Equimètre has their hands full trying to become a market leader. The ideal product is something that combines SeeHorse with EquinITy and not only allows the trainer to monitor the horse when it is training, but also after it is training and how long it takes to recover from a training session (via heart rate variability over the next 24 hrs) and what its long term health is.
Where Equimètre is different, and where I see a good gap in the market that may find them successful, is the use of machine learning and artificial intelligence programs to judge both the physical fitness, overall ability and alert for health issues of the horse and prevent injury. Getting data out of any of these devices is easy enough, but interpreting them is totally different and a lot of the interpretation offered isn't based on any real science (data or veterinary). While some of the products listed above give good displays of the data, they leave it up to the trainer to make an assessment of its worth. I think that if trained properly a good AI/Machine Learning algorithm would provide a lot more robust analysis of the data over time.
Ultimately there is no doubt that the industry is going to end up with an injectable biosensor or nanosesor that is placed in the racehorse (probably in the chest muscle between the legs) that not only monitors vital signs 24 hrs a day (heart rate, heart rate variability, cortisol, glucose etc) but also has a GPS and gyro accelerometer allowing for motion efficiency analysis (Dorso-Ventral, Medio-Lateral, and Crania-Caudal) as well as speed and stride length, along with real time lactate analysis during exercise. Wrap this data around some A.I/Machine Learning and we are starting to talk about a system that wouldn't replace a trainer, but would certainly let them know what they had in terms of class and help with fitness and health overall.