(ANTIMEDIA) Depression, which affects 6.7 percent of the United States population, is a notoriously difficult disorder to diagnose. This leads to misdiagnosis and, ultimately, over-medication and people being prescribed medication even though the nature of their depression may not be fully understood (for example, general anxiety disorder, bipolar disorder, and other mood disorders are often lumped into the same category). But a new algorithm may be able to help with this problem.
SimSensei uses something called the ‘k-means algorithm,’ which was originally developed in 1967 to help with statistical analysis in machine learning and early data mining research. SimSensei uses it to detect speech patterns in patients that may have depression or post-traumatic stress syndrome. People suffering from depression often speak in a monotone voice with longer pauses. Often, these mannerisms and linguistic nuances are extremely subtle, and some patients have even conditioned themselves to hide outward signs of their depression.
SimSensei uses frequency analysis to help doctors assess patients in virtual interview rooms.
One of the first major applications of the program was to assist with an interview training program for veterans about to integrate back into civilian life.
The SimSensei team also wants to use the technology to help diagnose other disorders, such as Parkinson’s and schizophrenia. The project authors also speak of their more abstract goal of destigmatizing mental illness so more people seek help:
“This effort seeks to enable a new generation of clinical decision support tools and interactive virtual agent-based healthcare dissemination/delivery systems that are able to recognize and identify psychological distress from multimodal signals. These tools aim to provide military personnel and their families’ better awareness and access to care while reducing the stigma of seeking help.”
This article (Meet the Virtual Robot That is Helping People with Depression and PTSD) is free and open source. You have permission to republish this article under a Creative Commons license with attribution to Jake Anderson and theAntiMedia.org. Anti-Media Radio airs weeknights at 11pm Eastern/8pm Pacific. If you spot a typo, email edits@theantimedia.org.