Using AI to diagnose developmental language disorder

Rebecca Bright MBE was recently named one of the winners of Virgin Media Business’ Voom 2018 competition for her business, Therapy Box. We caught up with her to find out more about the business and how they’re using technology in a unique way to benefit people with speech and language problems…

How did you come up with the idea for Therapy Box?

I trained as a speech and language therapist in Australia. After graduating in 2001, I went on to work in the healthcare system in Australia before moving to London in 2005. I worked in the NHS specialising in helping people with communication problems after strokes, brain injuries and as a result of other neurological conditions.

In 2010 I noticed that the communication aids that were on the market at the time were expensive and cumbersome. People who needed technology to support their communication were waiting for years for funding to come through to purchase machines that cost between £5000-12000. Then it was not uncommon for these devices to languish on shelves due to their heavy and outdated nature. My husband and I wanted to look at how we could create an iPhone app to help people with motor neurone disease to be able to communicate once they lost their speech.

Using my experience and personal motivation given my grandmother had passed away due to the disease, we designed the first version of the app, Predictable while still working in our regular jobs. The app launched in January 2011 and went on to be a big success. It is now in 11 languages and used by over 200,000 people. 

How is Therapy Box using technology to make speech therapists’ lives easier?

Speech and language therapy assessment involves recording and analysing patient speech. Language analysis norms have been established, but tests to assess this require manual transcription of speech, word counts and calculations. This detailed transcription and analysis of language is essential for determining effective intervention strategies and measuring outcomes for patients with speech and language problems.

However, this manual method is very time consuming and susceptible to human error. A language assessment which takes 30 minutes to carry out could take another 90 minutes of administrative work (transcription and analysis) in the background. Pressure on NHS speech and language therapists’ time due to increasing caseload sizes has cut the time available for this activity, which is vital for determining the correct intervention plan. This can lead to poor outcomes for children and an increase in speech and language therapist time needed to address the problem. A tool which can complete language transcription and analysis automatically would significantly reduce the time required by a therapist to complete this specialist activity. This would lead to better informed therapist management plans, improved outcomes for children and increased efficiency.

 What’s the next tech innovation that you would love to see developed?

We are working hard on our ATLAS project. Funded by NIHR i4i Connect, ATLAS is an AI-based platform, which listens to a child talking while they are presented with fun games and images on a smartphone. It then analyses whether they are at risk of developmental language disorder.

Our focus is on developing ATLAS and ensuring all our projects continue to reach the patients who benefit from them.

How will winning Voom impact your business?

Winning Voom has been a big boost to our team. We have appreciated the opportunity to gather feedback from the industry and get insights from other companies that may be further on in their trajectory in the AI healthcare space.

Comment

Our Companies

Quick Links