Evaluating the performance of the cognitive workload model with subjective endorsement in addition to EEG

Yamini Gogna,Sheela Tiwari,Rajesh Singla

DOI: https://doi.org/10.1007/s11517-024-03049-4

2024-03-05

Medical & Biological Engineering & Computing

Abstract:The aptitude-oriented exercises from almost all domains impose cognitive load on their operators. Evaluating such load poses several challenges owing to many factors like measurement mode and complexity, nature of the load, overloading conditions, etc. Nevertheless, the physiological measurement of a specific genre of cognitive load and subjective measurement have not been reported along with each other. In this study, the electroencephalography (EEG)-driven machine learning (Support Vector Machine (SVM)) model is sought along with the support of NASA's Task Load Index (NASA-TLX) rating scale for a novel purpose in workload exploration of operators. The Cognitive Load Theory (CLT) was used as the foundation to design the intrinsic stimulus (Spot the Difference task), as most workloads operators are exposed to are notably intrinsic. The SVM-based three-level classification accuracy ranged from 85.4 to 97.4% ( p < 0.05), and the NASA-TLX-based three-level classification accuracy ranged from 88.33 to 97.33%. The t- test results show that the neurometric indices contributing to the classification significantly differed ( p < 0.05) for every level. The NASA-TLX scale was utilised for validation in its basic form after the validity (Pearson correlation coefficients 0.338 to 0.805 ( p < 0.05)) and reliability (Cronbach's α = 0.753) test. This modeling is beneficial to phase out particular-level cognitive exercises from the curriculum during under or overload workload (critical) conditions.

engineering, biomedical,computer science, interdisciplinary applications,mathematical & computational biology,medical informatics