Demonstration of a Method for Real-time Detection of Anomalies in Team Communication

Abstract

Real-time analysis of team communication data to detect anomalies and/or perturbations in the team environment is an ideal method to improve on teams’ interactions and responses to potential crises. In this paper, we demonstrate a method to detect anomalies through observing communication patterns of neurosurgery teams. We simulated the real-time process by analyzing previously collected communication data to assess the effectiveness of a nonlinear prediction model to detect anomalies. We compared predicted values of communication determinism (a measure of how organized communication patterns are) to previous values in each team’s time series. These deviations formed a separate root mean square error (RMSE) time series, and we examined the magnitudes of the RMSE time series at the points of known perturbations. Additionally, we examined the effect of window size on perturbation detection. We found that our nonlinear prediction model accurately detected the perturbations and shows promise for future real-time analysis.

Neuroergonomics: Quantitative Modeling of Individual, Shared, and Team Neurodynamic Information

“…Developing more dynamic descriptions and measures of team macrocognition, situation awareness, and team coordination still remains a challenge, although progress in analyzing the dynamics of team macrocognition (Wiltshire, Butner, & Fiore, 2017) and situation awareness are emerging using communication (Gorman, Amazeen, & Cooke, 2010;Grimm et al, 2017;Kitchin & Barber, 2017).…”

Team Communication Patterns in Critical Situations

“…One explanation for this is that critical situations, by definition, are unique, ambiguous, emergent situations (Stachowski et al, 2009). Our study has provided valuable insights into the adaptation of communication patterns but more research on the topic is warranted, particularly given the possibilities of providing real-time feedback on communication patterns to teams in training (e.g., Gorman et al, 2020; Grimm et al, 2017; Kiekel et al, 2002).…”

“…However, this is still cross-sectional data collected at multiple time points (Maupin et al, 2020), which is insensitive to more fine-grained patterns of communication such as who initiates more interactions or to what degree team members are prone to address recently contacted team members. Other methods are sensitive to time but give limited insight into patterns: recurrence analysis (Gorman et al, 2012; Grimm et al, 2017) only measures determinism, and flow analysis of utterance events in teams (FAUCET) does not characterize specific patterns (Weil et al, 2008). Discourse analysis is time-sensitive, but not context independent (Maupin et al, 2020), limiting generalizability and comparison over contexts and teams.…”

Measuring Real-Time Team Cognition During Team Training

“…This resulted in a moving window %DET time series that was 120 s shorter than the original turn-taking time series (there were 60 zero-padded observations at the beginning and 60 observations truncated from the end due to the centered windowing procedure). Previous research revealed the method to be robust across a range of window sizes but that window sizes smaller than the size used here tend to reduce the temporal precision of perturbation detection (Grimm et al, 2017).…”