Deep Learning-Based Architecture for Social Anxiety Diagnosis

Abstract

Useful insights about social anxiety are drawn with the help of image data obtained from the patient’s smartphone. Hand shivering and negative thinking are common symptoms of social anxiety. In this paper, the architecture is provided for the diagnosis of social anxiety, which uses the patient’s smartphone image database to obtain hand shivering and negative thinking patterns. In the proposed Hand Shivering Pattern Analysis (HSPA) algorithm, a hybrid approach, Simplified-Fast-Alexnet (SFA) learning model is used to identify the images with motion blur, and then amount of blurriness in those motion blur images is calculated with the help of Variance of Laplacian method which gives the intense of the disease. The proposed Negative Sentiment Pattern Analysis (NSPA) algorithm finds out the negative thinking pattern using the text in the shared images. Statistical measures obtained using the proposed algorithms in this research work support the psychologist for a better understanding of the individual’s level of social anxiety.