A new algorithm that uses speech patterns to diagnose depression has been developed.
People with depression usually speak in a particular manner – their speech may become more monotone and they may start using longer pauses. That’s why scientists are developing a machine called SimSensei. This machine uses an algorithm that places large data sets into groups or patterns based on average values which can then be compared against ‘normal’ speech patterns. In other words, SimSensei uses an algorithm to find vowel sounds that are associated with depression.
Researchers used the new algorithm on more than 250 volunteers and asked them to fill out a self-assessment questionnaire. This experiment showed a significantly reduced vowel space in people that scored positively on the questionnaires (people who self-reported symptoms of depression). These findings could support treatment of affective disorders, like depression and even PTSD in the future.
However, the new algorithm was designed to work alongside doctors, not to replace them. In other words, SimSensei would be an excellent digital assistant that works alongside doctors as they assess patients.