MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) researchers developed a neural-network model that learns speech patterns indicative of depression from text and audio data of clinical interviews, which could power mobile apps that monitor text and voice for mental illness.