Dystech using artificial intelligence to help speed screening for learning disorders

The app can be used to detect disabilities such as dyslexia and dysgraphia in 10 minutes.

reading-a-book-speed-read-flickr.jpg

French-born Hugo Richard was diagnosed with dyslexia and dysgraphia at the age of 12, but he didn't look into reading and writing disorders until he established Dystech.

Dystech is an Australian startup that was co-founded by Richards, Mathieu Serrurier, Jim Radford, and Gilles Richard, and is responsible for developing a screening app for early detection of learning disorders, such as dyslexia and dysgraphia.

Speaking to ZDNet, Richard said through research, the company identified that assessing something like dyslexia is often tricky. He explained that having a reading difficulty can often mean many things and so, the diagnosis of someone with dyslexia is not always accurate, nor is there sufficient support for those who do have it.

"We've observed there's a big imbalance between supply and demand. There are many more individuals with dyslexia than there are providers who can provide a diagnosis," Richard told ZDNet.

"What that leads to is if you're a parent and you think your child has difficulty with reading, to go through a diagnosis you need to make an appointment with an educational psychologist. Chances are you have to wait at least six months to a year. The cost is not hundreds of dollars, it's thousands; it's just very expensive and very inaccessible."

The app, which leverages artificial intelligence (AI) and machine learning and has been built on Amazon Web Services (AWS) infrastructure, takes users through a 10-minute screening test. The test involves users reading aloud words that appear on the screen while being recorded using their smart device. Users are then informed about their likelihood of having dyslexia. 

The app uses datasets of audio recording from both dyslexic and non-dyslexic adults and children to train the AI.

"We've discovered there is some correlation between reading reaction time. It seems that people with dyslexia have a longer reading reaction time than people without dyslexia. The reading reaction time is the time between the individual tries to start reading the word and the time when the word appears on the screen," Richard said.

A similar assessment is used to screen someone for dysgraphia, but instead of using audio recordings, the AI technology assesses a person based on a photo of a handwritten text.

See also: How an award-winning AI-powered software is helping students with remote learning (TechRepublic)

Dystech is continually collecting data to improve the algorithm behind the app, Richard said.

"It can only become better. What will happen every couple of months is we retrain the algorithm with the new data that we've collected and very easily replace the new predictor with the old one directly under the AWS hat," he said.

The challenge for Dystech, however, is being able to access the data it needs. 

"When your child has dyslexia it's kind of private … [so] it's been challenging for us to obtain data. We do have an ethic approval but even then, there's a huge roadblock to acquiring it," Richard said. 

Despite these hurdles, Richard said he was optimistic that the Dystech app could potentially replace the current diagnosis process, hinting there are currently ongoing discussions to introduce Dystech to the Victorian justice system.

"Dyslexia has a large impact on life and often people who commit crime have some literacy problem, and dyslexia is the most common learning disability. The thing is you cannot pay AU$2,000 per person in jail [to get screened], it's just too expensive and the government doesn't have that type of budget," Richard said.

"What's happening is they've become interested ... they don't really need a full diagnosis, what they need is, 'Should we put them in a class to support them with reading and writing?' Because we're a very different business model than a traditional diagnosis process. We're much cheaper, more accessible, and we're instant, so you can get instant result." 

Related Coverage

Australian government sinks AU$19 million into AI health research projects

The funding will be divvied up between five research projects.

La Trobe University uses AI to bring mental health care to cancer patients

In a bid to provide personalised support to patients undergoing cancer treatment.

Muru Music looks to launch AI music therapy platform

Using music as a brain stimulant for the ageing population.

Australian government fronts up $19 million for digital health tech development

Elsewhere, RMIT Online and the Digital Health Cooperative Research Centre have announced a suite of digital health short courses.