Entropy for team communication pattern recognition.

This work aims to demonstrate entropy as a method that can be used to analyze team coordination using naturalistic team communication content. Much of team coordination happens through communication; understanding team communication is vital to better formulate and train teams for successful performance outcomes. Several decades of team communication research have led to various methods for analyzing team communication patterns. Many existing methods for team communication analysis have not been tested with naturalistic communication or only consider the frequency or flow of communication. Sliding-window entropy is used to analyze team coordination dynamics with team communication as a proxy. The resulting time series are evaluated using nonlinear dynamical systems analysis and clustering. Various team coordination patterns are identified using communication entropy at the team level. Entropy can be used to identify team communication patterns and their relation to team performance. While team coordination happens at the team level, a posteriori analysis indicates members' individual characteristics affect the overall team coordination patterns. In teams where the contribution is unequal, some members disproportionately affect the overall team coordination, which may undermine the team impact and affect team performance.