When the black dog is staring you in the face
UC Associate Professor Roland Goecke has worked with the Black Dog institute in Sydney to develop a computer program that uses voice and facial recognition to diagnose depression. The thermal camera face image analyses. Photo: Melissa Adams
Sufferers of depression may soon be monitoring the severity of their condition using smartphones and digital tablets incorporating voice and facial recognition technology developed at the University of Canberra.
Researchers say the computer program, created in partnership with the mood disorder treatment and prevention organisation the Black Dog Institute, has brought them closer to an objective indicator of melancholic, or biologically determined, depression.
The University of Canberra's Roland Goecke, an expert on human-computer interaction, worked with the institute to develop the program, which measures the indicators of melancholic depression, including a lack of facial movement, talking slowly, slumped posture and avoiding eye contact.
Using the program, a computer monitors the patient while they view images and video that commonly elicit certain emotional responses, analyses their voices during interviews and delivers a diagnosis based on the patient's reactions.
''One of the signs and symptoms of depression is that kind of interpersonal functioning is disturbed, and we are basically looking for those signs in an automated way,'' Dr Goecke said.
Institute researcher and clinician Gordon Parker said initial clinical trials yielded a 90 per cent accuracy rate, whereas existing measures achieved an average of 65 per cent accuracy.
Professor Parker said technology would help doctors determine an appropriate treatment program for patients. ''By and large, melancholic depression needs physical treatments like antidepressive drugs, and many of the non-melancholic disorders don't need drugs,'' he said.
It was difficult for patients with depression to objectively measure their symptoms, so using the program on their smartphone or other devices may help them identify when they needed to seek help, he said. It could also give doctors a clear indication of how patients were faring between appointments.
A clinical trial at the Black Dog Institute in Sydney tested the technology on 30 people suffering from depression, and compared the results with a control group of 30 people. It found that the voice and face recognition program gave the same diagnosis as two independent clinicians in 90 per cent of cases.
Stage two of the program will involve the researchers developing a prototype that will be distributed to psychologists and general practitioners to assist with their initial diagnoses of depression.
The third stage of the project will be developing the technology as a tool for patients to monitor their progress during treatment, on smartphones or other electronic devices.
Psychologists at the Black Dog Institute approached Dr Goecke about six years ago, looking to develop a more objective way of determining if a patient has melancholic, or biologically determined, depression.