UEL leads the pack in AI assistance dog innovation

A pioneering study led by the University of East London (UEL) is using artificial intelligence (AI) to predict which puppies are most likely to succeed as assistance dogs – offering a powerful new tool to support trainers and improve outcomes for people who rely on these highly skilled animals.

The project, headed by Dr Mohammad Hossein Amirhosseini, Associate Professor in Computer Science and Digital Technologies at UEL, applies machine learning and deep learning models to behavioural data collected during early puppyhood. The findings could significantly reduce the emotional and financial costs of dogs being released from training programmes early by identifying the strongest candidates earlier in the process.

“One of the biggest challenges in assistance dog training is the emotional and financial cost of late-stage failure,” said Dr Amirhosseini. “This is more than a tech innovation - it’s a leap forward for animal welfare.”

The research focused on behavioural questionnaires filled out by trainers when the puppies were six and 12 months old, providing snapshots of temperament, focus, and personality. AI models trained on this data were able to detect patterns and predictors of success that even experienced trainers might miss.

“This is a truly interdisciplinary project with real-world impact,” added Dr Amirhosseini from the School of Architecture, Computing and Engineering. 

We worked closely with organisations that train and place assistance dogs every day. These are the people who know how high the stakes are - not just for the dogs, but for the people whose lives depend on them.”

The study, published in the 50th anniversary issue of Applied Animal Behaviour Science, builds on previous UEL-led research, published in Nature, which used AI to map dog personality traits. This new phase takes that work a step further, using predictive modelling to support real-world decision-making for dog training programmes.

UEL collaborated with a global team of researchers and industry leaders on the project, including the University of Pennsylvania, the University of Arizona, the University of Nottingham, and several major assistance dog organisations. Among these were The Seeing Eye – the world’s oldest guide dog training centre – Canine Companions®, featured in Netflix’s Inside the Mind of a Dog – and Dogvatar, which funded the research and developed the core technology.

“We’re applying cutting-edge AI to one of the most human-centred roles animals can serve in – helping people live more independent lives,” said Dr Amirhosseini. “And we’re proud to be walking this path with some of the best minds in the field – both human and canine.”

Read the full study: Prediction of Assistance Dog Training Outcomes Using Machine Learning and Deep Learning Models.

Photo credit: Matt Seymour for Unsplash