Background
Glue ear (otitis media with effusion; OME) happens when fluid collects behind the eardrum, making it harder for sound to travel into the ear. It is the leading cause of hearing loss and disability in children, with around 80 percent having glue ear by the time they reach primary school. Glue ear is responsible for approximately 213,000 primary care consultations each year. Whilst glue ear is often temporary, some cases can last years if untreated. Hearing loss can negatively impact on health, wellbeing, education, and speech development, so it is important that glue ear is diagnosed and treated in a timely manner.
Glue ear is diagnosed by looking for the presence of fluid behind the eardrum (otoscopy) and measuring how the eardrum moves (tympanometry), but it can be difficult to diagnose. A missed or delayed diagnosis can lead to over-prescription of medication e.g. antibiotics (resulting in antibiotic resistance), and over-referral to specialist services e.g. ear, nose and throat specialist clinics (resulting in longer waits). If it was possible to improve the way a diagnosis of glue ear is made, then more children would get a correct diagnosis the first time they seek medical advice in primary care.
We will look to see if we can use a computer programme (a type of Artificial Intelligence, or AI) to help health professionals diagnose glue ear. We aim to recruit 500 children at our hospital Trust: 250 with suspected glue ear, and 250 without. We will use a smartphone to take a video of their eardrums (video-otoscopy) to look for fluid, and measure the pressure in their ears using tympanometry. These videos and pressure measurements will be used to train the AI to tell the difference between ears that do, and do not, have glue ear. After training, the AI will be tested to see how good it is at telling the difference compared to human experts. As well as developing this AI, we want to know what children, parents and carers, and health professionals, think about using AI-technology to support how healthcare is managed and delivered. We will explore this with an established questionnaire (given to 500 parents/carers) and interviews (with up to 20 people).
This project has been designed in collaboration with children, parents and carers. Feedback from those with lived experience is integral to this study, and we will provide continuing opportunities to influence our research by creating a Patient Steering Group (PSG). We will ask the PSG to write a summary of our findings, and work with our Patient and Public Involvement Lead to advise on how best to share them.
Impact of ENT UK Foundation funding
The ENT UK Foundation Research Pump-Prime Grant supported the research team to evidence unmet need, establish a multi-disciplinary team of study coordinators and collaborators, and horizon-scan for further external funding opportunities.
To evidence unmet need the research team held interviews with children and young people, their parents and carers, audiologists and ENT specialists, and conducted a pilot questionnaire survey of GP trainees’ attitudes to artificial intelligence in healthcare. These stakeholders have helped to co-design the study, will influence the study as we progress, and will aid with dissemination of findings.
A research team including ENT specialists, audiologists, academic qualitative researchers, a paediatric Public Health specialist and an expert Patient and Public Involvement lead has been assembled. We have established an industry partnership with TympaHealth Ltd, who will provide devices and machine learning expertise. This multi-organisation, multi-professional team has been cited as a strength of the project in subsequent successful awards.
Although it is a new award scheme, the prestige attached to the ENT UK Foundation Research Pump-Prime Grant proved instrumental helping the team secure further research funding from charities, NHS England and the NIHR Manchester BRC.
Current Status
Ongoing
View more