“Is my cat happy or not?”, is a question that has been haunting owners of notoriously inscrutable felines since forever – but it may soon be answered thanks to a new app.
The prototype app, dubbed Tably (previously Happy or Not), was released by Sylvester.AI – a joint venture between Canadian artificial intelligence technology company Alta ML and The Bar G, a portfolio of companies involved in sectors ranging from sleep gadgets to ranching, to animal health – and it can use a photo of a cat to determine its level of happiness based on facial expressions.
The app was launched after the developers at AltaML discovered the Feline Grimace Scale (FGS), a scientifically-validated tool for assessing acute pain in cats based on changes in facial expression published in Scientific Reports and Nature in 2019. This led the team to toying with the idea of applying their expertise in machine learning to foster healthier and happier lives for cats.
“The big challenge with cats is that they don’t express when they’re in pain,” says AltaML’s vice president of product, Chris O’Brien. “They go and hide, whereas a dog will come in and whine and nuzzle you.”
Prior to the FGS, cats were considered to be delphic creatures that did not give cues to signal if in pain. The truth is that these cues do exist – they’re just extremely subtle.
“Depending on the cats’ muzzle, eyes, ears, whiskers, or head position, you can determine whether or not a cat is in pain,” says Paulo Steagall, lead author of the FGS publication and associate professor of veterinary anaesthesia and pain management at the University of Montreal. The FGS was created by Steagall and his team to be used by veterinarians, other veterinary technicians, and cat owners.
While the FGS is available for public use, Alta ML coupled it with machine learning to make it more accessible to the general public. The app picks up relevant points on a cats’ face, maps them out based on the FGS, and shoots out a reading – happy or not – along with a percentage. The percentage isn’t correlated to the level of happiness, but rather the level of confidence the machine has.
“With a high quality and full face front image of the cat, the accuracy is 97 per cent [with adults cats of most breeds], which we are tremendously happy with,” says the app’s senior product manager, Michelle Priest.
According to Alta ML’s own survey of pilot app users, cat owners mostly use it to know how a cat is doing after having undergone a surgery, or to understand why a cat is constantly hiding in the home.
“It could be used to decide if you should take the cat to the vet or even an end-of-life decision,” Priest says. “You might have had a cat for 18 years, but you don’t want it to suffer. We also had this one guy say he wanted to find out why his cat was howling at night.”
Steagall did not comment on the validity of the app as he was not involved in its creation, but he says that in any application of the FGS, context is everything. He uses a photo of his own cat taking a nap as an example.
“She hears some background noises and she has her eyes partially closed and her ears backwards,” he says. “So she would get a high score on the grimace scale, saying that you should probably give what we call rescue analgesia, but she was not in pain at all. She was just sleeping and paid attention to the background noise.”
Priest says that Alta ML is currently working on a final version of the app for both iOS and Android, even if there is no official launch date yet.
The company plans to make Tably become part of a veterinarian’s post-appointment follow-up process, and Alta ML and Sylvester.AI are currently working with veterinary medicine professionals to make it possible.
“It has kind of become its own entity,” Priest says. “We’re starting with cats, although I think this idea could work with multiple species and grimace scales,” Priest says.
Lire la suite: www.wired.co.uk