OPINION

DIAGNOSING DISEASES VIA SPEECH

You walk into your office, and your colleague at the next desk greets you... but something is not quite right. You ask what's wrong and they tell you they're suffering from a terrible migraine that was induced by a horrible family fight. Aha, you proclaim! You knew it from the moment you walked in. But did you or did his greeting... signal a problem? Can we use Voice to detect illness? We're not talking about capturing words like 'I feel sick' or 'I have a fever' but the actual vocal qualities of our voice. When we produce sounds and form words, there's a lot of energy and complexity that goes into that. Our lungs and vocal chords, our overall energy levels, our mood, our mental and emotional state, all contribute as to how our voice will sound and, obviously, all these factors are prone to illness. We speak differently depending on our mood, so imagine what happens when experiencing mental health issues. 

Health professionals have been using vocal analysis for diagnosis for years. The Diagnostic and Statistical Manual (DSM) of mental health has been using speech and language to diagnose mental illness for at least 50 years (e.g., the second DSM, published in 1968, has “talkativeness” and “accelerated speech” as two common symptoms of what was then called manic depression, now termed bipolar). But what about diagnosing something like heart disease, migraines, or Parkinson’s disease using only the sound of someone’s voice? Amazon is taking personalised medication to a whole different level... they filed for a patent in 2017 that would allow Alexa to determine whether you're sick and... sell you targeted products. The Canary Speech app has successfully completed FDA clinical trials and is ready to detect Alzeheimer's, PTS, and depression for suicide prevention, while researchers from the Polytechnic of Porto, School of Engineering, in Portugal have submitted a paper suggesting a methodology for early detection of Parkinson's using signal and speech processing techniques integrated with machine learning algorithms. Meanwhile, the pharmaceutical giant, Boehringer Ingelheim, is working on an app using speech recognition to detect warning signals for Schizophrenia or Alzheimer's dementia.

There remains a lot to be desired, but the technology is looking promising. The way we diagnose health issues is bound to change, and voice is going to play a significant role. At Behavioral Signals, through Joana Correia's research work on speech analysis for clinical applications, the machine learning team has been able to work on building deeper models for detecting depression from speech. Depression and anxiety have a significant economic impact; the estimated cost to the global economy is US$ 1 trillion per year in lost productivity, according to the World Health Organization.

FROM OUR BLOG

FROM OUR BLOG

Collecting Data Alone... Is Not Enough

More than ever before, we are able to capture and analyze data about the customer journey. Chatbots, AI-agent assistants and other AI-powered tools provide actionable insights in real-time for the customer service and product development teams. While conversational AI is growing more sophisticated and can provide key insights, there is still very much a human element to understanding how something makes a customer feel – whether positive or negative – and acting on that knowledge subjectively. There isn’t always a clear yes or no answer hidden in the data. Artificial intelligence is attempting to fill this gap, however. 

Read more >