In 2014, we wrote about the launch of Samsung’s Strategy and Innovation Center (SSIC)’s wellness platform. In 2016, we covered the introduction of the group’s multifunction megachip, ready for applications. This chip has a new application that makes it possible to detect a user’s mood simply by listening to them talk. Detecting how people feel is usually highly subjective. Being able to detect changes in mood early and reliably via biomarkers could be a huge boon to caregivers, especially those who work with patients suffering from anxiety or with people with Asperger’s. It would help them take appropriate action before a problem escalates.
Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Institute of Medical Engineering and Science (IMES) published a study in which they describe a potential method for mood detection and prediction. The scientists used the Samsung Simband to sense and display movement, heart rate, blood pressure, blood flow, and skin temperature. Using artificial intelligence, the wearable was able to tell if a conversation is happy, sad, or neutral based on a combination of speech patterns, vocal pitch, and vital signs.
The scientists used two algorithms, one to classify the overall conversation and the other to break the conversation into five-second segments. In the reported test results, the current version of the MIT CSAIL system classified the five-second intervals with 18 percent greater accuracy than chance, and 7.5 percent better than current subjective methods. According to the researchers, continued work is necessary with more data, potentially with additional wearables, in order to increase accuracy rates and to improve emotional granularity beyond the current “positive” or “negative” mood ratings.
