Researchers developed a machine learning model that uses neural responses to predict the success of songs with 97% accuracy. In their study, participants listened to a set of songs while their neurophysiologic responses were monitored, generating data that helped the machine learning model determine potential hits.