top of page
Search

AI analyses breathing and speech to track infection progression


Southampton researchers are developing an app to help spot signs of a respiratory infection getting worse.


Respiratory tract infections (RTIs) affect parts of the body involved in breathing. This includes the nose, throat, airways and lungs.


The new app will support remote monitoring of people with these illnesses by recording their coughs, breathing and speech. It uses machine learning to analyse these sounds and check if they need to see a doctor.


The project is funded by UK Research and Innovation (UKRI) and led by the University of Cambridge. In Southampton, the team includes medical researchers and the Institute of Sound & Vibration Research (ISVR).


Common cause of illness


RTIs range from the common cold to more serious conditions, such as Pneumonia. Most get better without treatment, but some patients may need to see a GP.


Prof Nick Francis is Head of School of Primary Care, Population Sciences and Medical Education at the University of Southampton. He is also part of the NIHR Southampton Biomedical Research Centre.


“RTIs are the most common cause of illness, resulting in around half of all antibiotic prescriptions,” Prof Francis explained. “Most people get better quickly, but we need to notice quickly when people are getting seriously ill. If we don’t, the effect on them and on healthcare services can be large.


“We have tests that help doctors identify patients who are more likely to need treatment. But these do not work well for every patient and are not useful for helping patients manage their own illness.”


There are some signs, such as breathing faster, that suggest a RTI is getting worse. The infection may also affect the sound of speech and trying to breathe when speaking.


Prof Francis added: “We hope to develop an app that will assess all these signs and give warning when someone with an RTI should see their doctor. It will also rate its own confidence in its prediction, which will help doctors trust machine learning in healthcare.”


AI in healthcare


The study has received a £590,000 grant from UKRI. It is titled RELOAD (REspiratory disease progression through LOngitudinal Audio Data machine learning).


Prof Anna Barney, from the university’s ISVR, said:


“Using acoustic data is an underexplored application of machine learning. The AI in this application will reassure patients when an infection is self-limiting and direct patients to a GP when it is not. This will help to ensure that GP appointments are available for those that need them.”


The UKRI funding was part of a £13 million investment to boost use of AI in healthcare. Another Southampton project received funding to improve the analysis and interpretation of newly discovered variations in people’s DNA.

bottom of page