As their master's thesis, Tim Pagrell, Applied Physics, and Jennie Berndtson, Biomedical Engineering, have developed an AI-based method that analyzes dogs’ breathing and could support veterinarians when assessing whether a dog has BOAS. Breathing problems are common in short-nosed dog breeds such as pugs and French bulldogs, and veterinarians currently need specialised training to carry out a formal BOAS assessment.
The method is based on AI models that the students themselves developed and trained to identify patterns in breathing sounds. For their work, Tim Pagrell and Jennie Berndtson have been awarded the Bert-Inge Hogsved Prize for Best Entrepreneurship 2026.
BOAS – Brachycephalic Obstructive Airway Syndrome – occurs when a dog’s airways are too narrow. This can make it difficult for the dog to breathe, especially during exercise, and in some cases surgery is required.
“Today veterinarians need specialised training to carry out a BOAS assessment. We wanted to explore whether analysing a dog’s breathing could support that process and make the assessment easier,” says Tim Pagrell.
Recordings from dogs used to test the model
As part of her master’s thesis, Jennie Berndtson recorded breathing sounds from five dogs undergoing a BOAS examination at a veterinary clinic. The recordings were made both while the dogs were at rest and after brief exercise. The material was used to test the model, which had been trained on previously collected data.
The system analyses audio files and searches for patterns in the dog’s breathing. Background noise is filtered out, and statistical methods are used to identify the parameters that best represent distinctive features of the breathing sounds.
“The idea was to see whether patterns could be detected in the breathing that might not always be obvious when simply listening,” says Jennie Berndtson.
In tests involving the five dogs included in the study, the model produced correct results in all cases.
An interest in dogs sparked the project
Both Tim Pagrell and Jennie Berndtson grew up with dogs, and Tim currently has two of his own. When they were looking for topics for their master’s theses, they therefore chose to work on a project related to dogs’ breathing.
“I usually try to choose projects that feel important to me. When I started reading about BOAS, I was quite surprised by how common the problem is in certain breeds. That made it feel even more important to work on in my thesis,” says Jennie Berndtson.
The goal is that veterinarians themselves would be able to record a dog’s breathing and have the model analyse the sound file.
More recordings could help develop the model
Tim and Jennie would like to continue developing the method, but that will require more recordings from dogs undergoing BOAS examinations.
“I would very much like to continue developing the model. It already works today, but more data from veterinarians would make it possible to improve it further,” says Tim Pagrell.
He also sees the possibility that the analysis could eventually be carried out directly through a mobile app.
During the project, the students were in contact with researchers at the Swedish University of Agricultural Sciences (SLU). The hope is that more veterinarians will become aware of the method and contribute recordings of dogs’ breathing so the model can be trained on more data. With more recordings, the method could be further developed and eventually become a practical tool in veterinary work.
The project is based on two master’s theses carried out at Chalmers University of Technology.
About the prize
The Bert-Inge Hogsved Prize for Best Entrepreneurship was established in 2011 by Bert-Inge Hogsved, physicist and founder of the Hogia Group. The prize is awarded annually by the Forum for Engineering Physics at Chalmers University of Technology to students in engineering physics, engineering mathematics or chemical engineering with physics. The aim is to encourage entrepreneurial initiatives among students.
The 2026 prize amounts to SEK 30,000 and is shared equally between the two recipients.