Emotions give important clues about a person's health. Tracking emotions over time can help measure health in the long run. This is especially important for people with bipolar disorder, where problems with emotion regulation signal worsening mood. Usually, measuring emotions requires people to assess themselves, which is outside their daily routine. In a recent paper, Prechter Program researchers introduce a new method to collect real-world speech data from daily life and measure emotions from this data. The team's approach combines a new way to collect data and reliable emotion recognition models. Researchers tested this method alongside traditional clinical and self-report mood measures. They show that both passive and self-reported emotion data help us accurately estimate mood severity in people with bipolar disorder.