Neural activities of the brain can be recorded using various techniques, being the electroencephalography (EEG) the most used due to its characteristics and accessibility. Brain-Computer Interfaces (BCIs) enable direct communication between the brain activity and external devices, benefiting areas such as rehabilitation. Among BCI paradigms, Movement-Related Cortical Potentials (MRCP) are employed to detect movement intention and control external devices. However, mental fatigue resulting from prolonged cognitive activities, can affect MRCP amplitude, making its detection more challenging. The aim of this study is to implement Machine Learning Methods to detect MRCP and evaluate the influence of mental fatigue on their characteristics. EEG data from healthy subjects with no prior BCI experience were used to extract features from 9 electrodes, which were analyzed using Support Vector Machine (SVM) and Naive Bayes (NB) to detect MRCP. The results showed that under initial trials conditions, MRCPs were more consistent and detectable. Under suspected fatigue due to overwork, MRCPs exhibited greater inconsistency, making their detection by ML models more difficult, which could indicate that mental fatigue negatively affects the ability of the models to detect MRCP.