Mental health is a major pillar in the overall health of a human being and allows the person to function well in society to fulfill his/her needs, responsibilities, and desires. Yet according to World Mental Health Report (WHO), more than one out of a hundred deaths is caused by suicide, making it a major cause of death among young individuals. With social media platforms serving as an easily accessible platform for a person of any demographic, age or sex, to voice their thoughts, and read unvalidated opinions about sensitive topics like mental health, boundaries between personal expression and potential threats to mental well-being are constantly becoming blurry. With a certain increase in online discussions regarding mental health among various individuals, it becomes important to explore these linguistic nuances of comments and analyze them for better understanding as well as the prediction of mental state of the individual. Using natural language processing techniques, this study aims to establish the patterns and sentiments conveyed by commentary on social media, to better identify potential indicators of mental health issues. Using a bidirectional representation natural learning model (BERT) and a simple three-layer architecture, a dataset containing over 27 thousand text entries, relating to people with anxiety, depression, and other mental health issues social media interactions, was analyzed. The model achieved an accuracy of 87.68% and 85.45% precision.