Machine Algorithm Can Predict Whether A Relationship Will Last Based On How Couples Talk To Each Other


A new piece of research claims it can predict how long a relationship will last based on how couples talk to each other.

Titled “Predicting couple therapy outcomes based on speech acoustic features”, the study was written by researchers from the University of Southern California and published in PLOS ONE.

The work studied 134 couples who were in relationship therapy, paying special attention to the conversations they had with each other.

The researchers wanted to find out whether the “acoustics of spoken interactions” of clinically distressed spouses could give hints as to whether the therapy would improve their relationship.

The team analysed the vocal intonation and intensity, pitch and tone of each individual, and fed it into a machine-learning algorithm.

This algorithm was trained to find a connection between these vocal features and predict whether the therapy would be successful.

Professional therapists were then asked to make their own predictions as to whether therapy would be successful based on videos of the couples’ counselling sessions.

Comparing the two, the team found that the machine-learning algorithm was “comparable or superior” to these experts in predicting whether a couple’s relationship would improve.

By focusing on the vocal characters, the algorithm was correct 79.3% of the time, compared to the experts who predicted correctly 75.6% of the time.

The University of Southern California researchers said: “This finding supports the notion that behavioral outcome, like many other behavioral aspects, is closely related to the dynamics and mutual influence of the interlocutors during their interaction and their resulting behavioral patterns.”

Read more here.