Exploiting the way speech is slurred, elongated, or spoken in a more nasal tone, allows a program built by a Boston based company to produce an initial diagnosis. The program can detect tiny anomalities in the voice which are not detectable to the human ear.