In a nuclear power plant (NPP), operator performance is a critical to ensure safe operation of the plant. The fitness for duty (FFD) of the operators should be systematically assessed before they engage in duties related to reactor operations. This study proposes the use of an electroencephalography (EEG)-based deep learning algorithm to classify an operator's FFD. To determine the suitability of this approach, EEG data were collected during simple cognitive exercises designed to examine the mental readiness of nuclear operators. The EEG-based FFD classification system designed could successfully determine an operator's sobriety, stress, and fatigue in a timely and cost-effective manner. As protecting personal information of the operators while using their EEG data is important and necessary, this study also investigated schemes for providing information security to the EEG-based FFD status classification system by following the International Organization for Standardization/International Electrotechnical Commission standard. Data confidentiality, integrity, and unlinkability were considered in the resulting schemes of information security for the EEG data. The resulting system provides the necessary protection of personal information and the FFD databases without significantly affecting the overhead of FFD classification through near real-time analysis.