Connect with us


Could speech processing improve mental health diagnosis?

Avatar photo



Speech technology company Intelligent Voice is looking to bring speech diagnostics into mental health care. Health Tech World heard from its CEO about the technology and how speech processing could be a valuable tool for supporting mental health diagnosis


The UK-based firm works with a range of organisations, in particular those that deal with sensitive information, including banks, the police and insurance companies. The technology can transcribe conversations and offer insight into what is being said.

Around 18 months ago, Nigel Cannings, CEO of Intelligent Voice, had a conversation with a call centre about issues it was having with burnout. At that moment, he realised that the technology they had developed could make a significant difference to people’s lives and their mental health.

“It was one of those lightbulb moments. It dawned on me that we were sitting on something here,” Canning told Health Tech World.

“Mental health is one of the more difficult things to diagnose and detect. There are different traits that people adopt [including] a change in [their] language.

“Over time, you might have two or three sessions with someone and they become less engaged, they’re speaking more slowly, the pitch variation is less, it becomes more flat and monotone.

“You might also see things like changes in pronouns and the way people talk; the way they begin to distance themselves from what’s going on around them.”

These changes in language can be a powerful indication of a mental health issue such as depression. While someone with bipolar disorder who is experiencing a manic episode may display what is known as pressured speech. This is where a patient speaks rapidly often in an agitated way, jumping from idea to idea and interrupting people at inappropriate times.

Latest HT World video interview

Intelligent Voice already provides a similar kind of behavioural analysis in other sectors, notably the insurance industry. The company uses various methods, including AI, to detect fraud by analysing patterns in speech.

Now, the company is repurposing these methods and applying them to mental health and wellbeing. The technology is in its early stages of development with plans to start testing the platform with a large public sector organisation in September.

The company’s team includes health experts in the fields and is consulting with mental health professionals, medical providers and clinicians to help build out the system.

This transition into the healthcare sector, Cannings said, has come with a number of challenges.

“One of the biggest challenges when you look at healthcare is the language. We spent a lot of time retraining a lot of our models to better understand the way in which a doctor would speak and then trying to detect certain things which are important in those contexts,” he said.

Cannings anticipates that patients might have concerns when sharing their voice data. Although most people are happy to have their emails stored in the cloud, the candid nature of a conversation with a doctor could lead to some feeling uneasy.

“People are very concerned about voice data; it’s a really interesting thing. As soon as people start talking about storing voice data in the cloud, they get much more nervous about it,” Cannings said.

“People might be worried about where that data is going to end up and who might have access to it. And then the thought that a robot might then be sitting there judging your mental state, it’s a bit scary, really.”

Cannings stressed that speech diagnostics in mental health care could not replace a clinician, rather it would act as a tool to support their diagnosis.

“We don’t talk about the decision being made by a machine, because it’s not, these things are indicators,” he said.

“It’s about helping the clinician with potential diagnostic information, but leaving the decision in the hands of the human. I think that’s where we have to get to with all of these technologies to make sure that people realise that the machine will not decide.”

The stigma surrounding mental health and the sensitivity of voice data could mean that the technology takes over 5 years to “take off”, Cannings said. However, if it is adopted, he believes the technology could improve the accuracy of diagnoses, help reduce NHS waiting lists and be used to monitor patients’ progress during their treatment.

“We could help people in the same way that an Apple Watch does,” he said. “It measures my beats per minute, it measures my EKG, my oxygenation, and I can track over a month, week or year whether my fitness is improving.

“We can use these types of tools to produce similar types of dashboards which would then alert someone on the other end. We prescribe vast amounts of antidepressants to people but we have no real way of measuring its efficacy. Imagine if you could automatically monitor whether someone’s mood was improving or not.”

Having been diagnosed with bipolar in his early thirties, Cannings has a personal interest in mental health. He recently talked about his condition publicly for the first time.

Working as a lawyer at the time, Cannings began experiencing severe mood swings. He took a trip to visit his GP and explained his symptoms, however his doctor was adamant that he was not bipolar and prescribed him a course of antidepressants.

“The [antidepressants] took my condition from being bad to being very bad, very rapidly […]I had been prescribed the wrong medication,” Cannings said.

He later went to see a private specialist who told him that his GP was wrong; he was “without doubt” bipolar and the medication he was taking was “extremely dangerous” for people with his condition.

“An antidepressant given to somebody who is bipolar can take the cap off the manic end. What can happen is you keep going up.”

While travelling on business, Cannings experienced what he called a “full blown” manic episode. He experienced hallucinations and was forced to arrange a flight home. Two days later, he was checking into The Priory, a private mental health care facility.

He was prescribed a medication that worked for him and he was able to recover from the episode.

“My life would have gone a very different way if my GP had just said ‘yes, you’re bipolar’,” Cannings added. “The whole terrible breakdown, which involved six months off work, would never have happened.

“I spent a long time blaming my doctor, but then I thought about it afterwards and realised that actually, this guy had 10 minutes, he had some bloke turning up saying he’s got a very serious mental illness and he’s not got many tools at his disposal.”

This is where Cannings sees speech diagnostics having the biggest impact; as a means to prevent incorrect diagnoses and support an underfunded mental health service.

His experience has also influenced the way he runs his company. Intelligent Voice has a programme that enables staff to confidentially seek counselling and get access to CBT.

“People in business need to talk about [mental health],” Cannings added. “I didn’t know a single person when I was going through this who had suffered as I had and who I could talk to about it with. Talking to someone would have made me feel a lot less alone.

“For those of us who run companies, we can put in place schemes whereby people who are feeling under pressure can talk to someone, be it a manager or an outside counselling service.

“We have to treat mental illness like a proper illness.”

Continue Reading


  1. Pingback: Intelligent Voice

  2. Pingback: Handheld eye test could screen for autism in children | Health Tech World

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending stories