Researchers from the University of California (Los Angeles) used an artificial intelligence system to analyze linguistic patterns and physiological signs to detect conflicts in couples. The work, which the team published in the IEEE Computer magazine, shows that it is the first time that this monitoring has been demonstrated to work outside a psychology office.
Project led by USC's Adela Timmons, researchers used smart phones and wearables on 34 couples in the comfort of their homes. The devices were equipped with machine learning algorithms that could overcome classic patterns of speech and physiology, such as an increase in heart rate and skin conductance levels, two signs that previous studies had shown to be associated with, the conflict.
During the day-to-day study, a sensor on each partner's chest measured heart rate, a wristband measured skin conductivity, and a smartphone collected audio recordings while the participants were tracked via GPS. The smartphone would ask couples to confirm if a conflict occurred when it was detected.
All things considered, the system managed to accurately detect conflicts 79,3% of the time. With language cues alone, the accuracy rate was a less impressive 62,3%. However, the system's ability to detect conflicts does not yet represent predicting or even preventing altercations.
With slow but sure steps, in the near future, we will therefore have an application that can prevent and prevent conflicts between couples.
