Possibility of disease discrimination by voice

Abstract

We developed indicators for mental health status using voice. It became possible to estimate the degree of stress and depression from the change of mental health status, but since there are many diseases that cause changes voice, the current algorithm cannot discriminate depression strictly. Therefore, we are working on developing a system that can distinguish various diseases by voice. Currently, we have found a voice feature specialized in major depression, which correlates with severity by doctor's diagnosis. Furthermore, we also found that voice features specialized in Parkinson's disease, and showed that both diseases can be distinguished by using these features. In addition, we showed that voice can be used for judging the therapeutic effect of sleep apnea syndrome, and for differential diagnosis of major depression and bipolar disorder. Now, we are studying the detection of dementia and are expanding the target diseases.