The use of machine learning methods in the diagnosis of depression - a systematic review

Abstract

Introduction: Nowadays, depression is one of the most common illnesses, affecting over 300 million people worldwide. Over the course of a lifetime, several dozen percent of the adult population of various ages and backgrounds will experience it. Depression is a debilitating disorder characterized by low mood, reduced interests, impaired cognitive functions, and vegetative symptoms such as sleep or appetite disturbances. According to the World Health Organization (WHO), it is the leading cause of disability and has a significant impact on quality of life, both privately and professionally. Its symptoms are heterogeneous and often overlap with other disorders, which means that this illness is not always accurately diagnosed. To assist physicians, in recent years researchers in this field have developed numerous machine learning methods aimed at improving the diagnostic and treatment process of this condition.

Aim of the study: The aim of this study is to summarize the current state of knowledge on the use of machine learning methods in the diagnosis of depression.

Methods and materials: A literature review was conducted using the PubMed database with relevant search terms.

Results: Articles describing studies that applied machine learning methods to predict and identify depression were included in the review. The machine learning models in these studies used data derived from neuroimaging, electronic health records (EHR), and peripheral blood transcriptomes.

Conclusion: Our review suggests that machine learning has great potential in the diagnosis of depression. This area of artificial intelligence offers new ways to analyze data and automate diagnostic processes. In the future, machine learning algorithms may become an integral part of standard psychiatric diagnostics.