Design and Evaluation of a Brain Signal-based Monitoring System for Differently-Abled People

The main goal of this paper is to design a system that will help actively monitor differently-abled patients in hospitals by capturing their brain signals which pass in the form of EEG signals. The signals are captured by observing the potential difference caused when an electric signal is passed during an instant time. Further, during active events, the changes are recorded and a range is assigned so that the values are mapped which would enable us to identify and recognize the state of a particular person. Later, this data is used to train a machine learning models which helps to classify a brain signal to a particular state that it resembles the most (using the pre-defined range as an output at the end). Further, tuning the model to improve generalization and hence concluding with the performance comparison of specialized machine learning algorithms to classify the input signals in terms of Accuracy, Precision, Recall, and F1 scores. This study also discusses about the challenges including the high level of noise in the EEG signals, which can significantly affect the accuracy and reliability of the data. Another challenge is the limited number of training examples, as collecting large amounts of EEG data from patients can be time-consuming and expensive. In future, the system can be enhanced by integrating smart IOT technologies such as sensors and buzzers to raise alerts to the concerned people.