Identification of Mind Wandering Associated with Different Workloads Using EEG in a Simulated Drilling Experiment

In high-pressure drilling environments, mind wandering is a significant contributor to unsafe behaviors and reduced task performance. Perceived workload plays a critical role in mitigating mind wandering, highlighting the need for an objective method to monitor workload-related mind wandering states among drilling crews. This study proposes a wearable EEG-based approach to identify and classify mind wandering, taking workload into account. A simulated drilling experiment involving 50 participants was conducted to collect neurophysiological and subjective workload data. In our two-stage analysis approach, a Gaussian mixture model (GMM) clustering method was used to identify mind wandering samples, and the NASA Task Load Index (NASA-TLX) scale scores were used to label the samples' perceived workloads. Second, a supervised learning algorithm was applied to classify the mind wandering associated with workloads. The results showed that the neurophysiological responses to mind wandering differed across workloads. The algorithm classification performance analysis showed that Support Vector Machines (SVMs) demonstrated the best performance in both levels of classification, with 90.06% for classifying centralized/mind wandering (1-Level classification) and 73.56% for identifying mind wandering associated with high workloads (2-Level classification). Overall, this study demonstrated the feasibility of applying wearable EEG devices to identify and classify mind wandering in drilling crews.