Locked-in Syndrome patients are often misdiagnosed and face pessimistic prognosis because of similarities with disorders of consciousness, a lack of objective biomarkers and a difficult-to-recognize pathogenesis. Biomarkers show promise in identifying similar conditions, utilizing electroencephalography (EEG) data. This data, particularly in the form of event-related potentials (ERPs), while successful in varying applications, suffers from methodological constraints and interpretation obstacles. The study documented in this body of work explores a machine learning paradigm with regards to N400 ERP data retrieved from a sentence comprehension task to tackle these hindrances and proposes a new auxiliary diagnostic tool for LIS and possibly disorders of consciousness. A support vector machine (SVC) and a random forest classifier (RF) were able to classify conscious individuals from unconscious ones with optimistic performance metrics. Based on these results, the proposed models and continuations thereof present valuable opportunities for the development of an auxiliary diagnostic tool for the classification of LIS patients, aiding diagnosis, improving prognosis, stimulating recovery and reducing mortality rates.