Machine Learning Techniques for Anxiety Disorder

Abstract

In recent years, artificial intelligence based applications have been improved and used to improve the timing, sensitivity and quality of diagnosis of psychiatric diseases. We aim to review the existing literature on the use of artificial intelligence techniques in the assessment of subjects with anxiety disorder. We searched databases about DSM-5 (Diagnostic and Statistical Manual of Mental Disorders) one of the main categories of anxiety disorders; Separation Anxiety Disorder, Generalized Anxiety Disorder, Panic Disorder and Social Anxiety Disorder between 2015-2021. We identified 30 different techniques on these works. Comparisons have been made with more than one algorithm in the studies. The Random Forest Algorithm has been seen in the most used machine learning method among these algorithms. In addition, the best accuracy performance has been observed in the Random Forest Algorithm. This article critically analyzes these recent research studies on anxiety. Considering the clinical heterogeneity of the data obtained from anxiety patients, we conclude that artificial intelligence techniques can provide important information to clinicians and researchers in areas such as diagnosis, personalized treatment, and prognosis.