A decision modeling approach for smart training environment with motor Imagery-based brain computer interface under neutrosophic cubic fuzzy set

Abstract

The modeling of smart training environments (STEs) for motor imagery-based brain–computer interface (MI-BCI) falls under the multi-attribute decision analysis (MADA) due to three main concerns, namely, multiple evaluation attributes, data variation, and attribute prioritization. Despite the tremendous efforts over the last years, none of the developed STEs have met all of the essential smart attributes. Thus, modeling multiple STEs to determine the best one for MI-BCI is difficult. Literature reviews have evaluated and modeled the existing STE alternatives, but informational uncertainty remains an open issue. The earlier MADA solution also has some issues. Thus, this study extended fuzzy weighted with zero inconsistency (FWZIC) with neutrosophic cubic sets (NCSs) for modeling uncertainty to prioritize the smart attributes of STEs and estimate the weight values of each one. Then, the developed NCS–FWZIC method is integrated with multi-attributive border approximation area comparison (MABAC) method to model the STE alternatives. The findings revealed the following: (1) NCS–FWZIC had effectively prioritized and weighted the smart attributes of STEs with no inconsistency. Ease of use attribute was considered the most influential attribute because it earned the greatest weight value. (2) MABAC method produced stable and reliable modeling results. STE5 obtained the highest model among the 27 STEs. Sensitivity analysis and Spearman's rho, systematic modeling, and comparison analysis were conducted to test the stability and robustness of the results reported in this study.