Brain-computer interfaces allow individuals with disabilities to communicate using only their thoughts. However, long-term BCI performance is often compromised by changes in the mental state of the BCI user. For instance, users often experience changes in fatigue, frustration, and distraction during BCI operation. However, BCIs do not have any way to react to these changes appropriately, and, consequently, exhibit decreased performance when they occur. In this study, we are investigating ways to detect these underlying changes in mental state in real-time and develop a BCI that can adapt to them to ensure that high performance is maintained over long periods of time.