Dr. Dolittle, meet AI: new model identifies pain in cats

Scientists use deep learning models and facial recognition analysis to detect subtle changes in ear tips, eyes and whiskers that indicate pain.

Naama Barak writes about a fascinating discovery that will help cats and their owners. Great news!

As any cat owner knows, felines can be pretty difficult to read – a problematic situation when our beloved pets are in pain. But now, a new artificial intelligence model can help us identify when our little tigers need that extra TLC. The reason it is difficult to tell when cats are in pain is their evolutionary ability to hide discomfort. This can result in house cats suffering from chronic pain without their humans ever knowing about it or taking them for treatment.

A team of researchers from the University of Haifa’s the Tech4Animals lab is developing the Dr. Dolittl-E app to detect animal feelings, using AI technology to advance and transform the world of veterinary care.

Along with researchers from São Paulo University in Brazil and Lincoln University and Nottingham University in the UK, they used deep learning models and facial recognition analysis to achieve a success rate of more than 70 percent in identifying cats in pain.

Their research, recently published in the Scientific Reports journal, is based on a study of 29 British shorthair cats. The researchers photographed female cats’ faces before and after sterilization, while they were still under the influence of painkillers and after the painkillers had worn off.

The AI models detected subtle changes in facial expressions, for example in the tips of the ears, eyes, whiskers and mustaches, that indicate pain. The areas around the mouth and eyes were found to be most significant in identifying pain.

The researchers say that their findings could change the way in which we care for cats, enabling anyone to photograph them in order to know whether they are in pain – and without the need for physical contact.