CNNs for Anxiety Risk Classification and Prediction in Mental Health

Anxiety, a mental health disorder that plagues many college students, is generally defined by persistent worry and fear. Early identification and accurate anxiety level assessment are also important for effective intervention. This study used a convolution neural network model (CNN), which aims to classify the students’ anxiety level into four categories, i.e., mild, moderate-severe and minimal. Utilizing CNN’s capacity to identify spatial patterns in data, the model was able to detect spatial patterns, and thus achieved low Mean Squared Error (MSE) of 0.02 with a desirable accuracy or 98%. The CNN’s architecture effectively identified small differences in symptoms that are frequently associated with the demands of academic life, including anxiety, excessive worry, restlessness, difficulties relaxing, and irritability. This reveals a massive scope for CNNs to be employed as mental health diagnostics at universities, providing them with an effective tool that allows personalized interventions and better anxiety risk assessment resulting in the overall well being of students.