DETERMINING TEAM COGNITION FROM DELAY ANALYSIS USING CROSS RECURRENCE PLOT Nasim Hajari, Irene Cheng, Bin Zheng and Anup Basu Department of Computing Science Department of Surgery University of Alberta ABSTRACT Team cognition is an important factor in evaluating and deter- mining team performance. Forming a team with good shared cognition is even more crucial for laparoscopic surgery ap- plications. In this study, we analyzed the eye tracking data of two surgeons during a laparoscopic simulation operation, then performed Cross Recurrence Analysis (CRA) on the recorded data to study the delay behaviour for good performer and poor performer teams. Dual eye tracking data for twenty two dyad teams were recorded during a laparoscopic task and then the teams were divided into good performer and poor performer teams based on the task times. Eventually we studied the de- lay between two team members for good and poor performer teams. The results indicated that the good performer teams show a smaller delay comparing to poor performer teams. This study is compatible with gaze overlap analysis between team members and therefore it is a good evidence of shared cognition between team members. 1. INTRODUCTION Eye tracking technique, as an objective assessment of surgical skill has been well documented in the literature [1]. Gaze pat- terns have been shown to differentiate poor and elite surgeons in several studies [2, 3, 4]. Also eye tracking can be used to examine the workload and vigilance of surgeons [5, 6]. Video analysis of an endoscopic cutting task performed by one vs. two operators indicates that good team collaboration results in superior team performance [7], and higher frequency of anticipatory movement was noticed in dyad teams [8]. Later on, Khan and Zheng [3] used dual eye-tracking to examine the spatial similarity in eye-tracking between two surgeons. Reporting level of gaze overlap is an innovative step in the study of shared cognition between two surgeons in a laparo- scopic team. However, people may gaze on the same spot at different time slots. To further analyze the similarity of eye tracking, we need to take temporal features into consideration in addition to spatial ones. In 2005, Richardson and Dale first used CRP to analyze gaze similarity recorded from two different persons [9]. They studied the relationship between a speaker and a listener based on their eye movements, and found that the coupling between a speakers and a listeners eye movements indicate if the listener was engaged to the speaker or not. While the gaze movement of the speaker was recorded, he watched a television show and at the same time talked about it. Later, the listener watched the same show as he was listening to the previously recorded monologues and his gaze move- ments was recorded too. Finally CRA was used to detect the matching behavior between speaker and listeners gaze movement. Marwan et al. presented a comprehensive review on different CRP and CRA approaches [10]. One can find an excellent MATLAB toolbox or an R package [11] to per- form CRP analysis. CRP can be used to study the differences between two processes or for the alignment and search for the matching sequences of the two data series even when the cross correlation fails or if the system is dynamic. It is a major task to find relations or interrelations between the time series. Linear data analysis is not suitable to analyze the short, non-stationary and complex data series. An appropriate method to analyze this type of data is Recurrence Plots (RP). It has been proven that recurrence is a fundamental property of dynamic systems, which means that after some time the system will reach the state that is arbitrary close to the former states and pass through a similar evolution. RP can visualize the recurrence behavior of dynamic systems. Also, one can perform the Recurrence Analysis (RA) based on the RP. CRP is an extension of RP. It can help in investigating the time dependent behavior of the two processes. The basic idea is to compare the phase space trajectories of two processes in the same phase space. CRP is used to study the similarity of the two different phase space trajectories. 2. BACKGROUND Cross Recurrence Plots (CRP) can be used to study the differ- ence between two processes or for the alignment and search for the matching sequences of the two data series even when the cross correlation fails or if the system is dynamic. It is a major task to find relations or interrelations between the 978-1-4577-0220-4/16/$31.00 ©2016 IEEE 3482