Depression and Suicide Risk Prediction Based on Machine Learning Models

Abstract

Nowadays depression is one of the most series diseases in the world. According to incomplete statistics, depression patients in the world are up to tens of millions of people. About 800,000 people committed suicide because of depression, most of them young people aged 15-29. In other words, on average, every 40 seconds someone commits suicide because of depression. In addition, most depressed patients have suicidal tendencies. Research of World Health Organization show that in the 10 years from 2005 to 2015, the total number of people suffering from depression increased by 18.4 percent. Anxiety and stress will be with human over the next 10 years and will become the most common mental illness, according to the World Health Organization. Therefore, pay more attention mental health. Addressing depression can alleviate problems such as healthcare costs that exceed national and corporate budgets. Using machine learning models to predict patients with depression is a key challenge in the field of clinical data analysis and is one of the prevalent techniques for predicting disease. In this paper, a deep network model for anonymous predictive analysis of the data sets provided by "Suicide Watch" and "Depression" on the Reddit platform and finally ensured that the accuracy of each model result reached 95%. Finally, the causative factors of depression were analyzed to help patients and medical staff to prevent depression in time. So that mental health and safety issues are paid attention to reduce the health and safety problems caused by depression.