Is there any way to Identify Depression on Twitter?

Photo by Milada Vigerova on Unsplash

Mental health continues to be a prominent plague for the civilized world. According to the current findings, one in four American citizens suffers from a diagnosable mental disorder in any given year. This not only happens in the USA but in India also depression has become a major disorder. In every country, there are certain numbers of depression patients can be found. But sometimes people don’t know that they have depression, in this article I’m going to talk about clinical depression which is known as Major Depressive Disorder. Nearly 300 million people worldwide suffer from clinical depression.

One in four cancer patients experience depression, one in three heart attack survivors undergo depression, and up to 75% of individuals diagnosed with an eating disorder will encounter the disease.

There can be several reasons behind the depression, and those reasons vary with each other. If we take one person from the USA and another person from India and try to compare the facts why they got depression those things can be totally different from each other, and sometimes there can be common reasons as well but most of the times there’s a huge possibility that reasons can be totally opposite. It’s because of the sociodemographic factors, as we know USA citizens already used to have a comfortable life compared to most other countries, but India which is an Asian country those people have to endure so many difficulties throughout their whole life. Because India has the highest population and it is still a developing country, so there are so many obstacles to citizens when they are going to find a job, entering a state university, etc. Not only in India as you can see in most of the Asian countries have to face those difficulties. I take the USA and India as examples because those are well-known countries by everyone, most of you have an idea about the difference between the lifestyle of people who are living in those countries. Depression can lead anyone to suicidal actions as well, so that is why we should pay attention to this disorder because it is a serious issue.

In this article, I’m looking forward to talking about how to identify depression on Twitter because people increasingly utilize social media platforms to share their innermost thoughts, desires, and voice their opinion on social matters and Twitter is one of the most popular social media that people used to share their thoughts and opinions. There are so many researches already done and researchers have been recognized that Twitter users with depression through questionnaires, surveys, and patient’s self-reported experiences. Most of them were collected Twitter data by using a crowdsourcing technique. Center of Epidemiologic Studies Depression Scale (CES-D), Beck’s Depression Scale(BDI), and Zung’s Self-Rating Scale (SDS) are commonly used by researchers in order to identify users with depression and get tweets from them with the aim of train their models. Sometimes they used tweets from those who are already mentioned in their Twitter bio that they diagnosed with depression. Tweets can be basically analyzed by using text, images, and emojis. But there can be so many problems when we analyzed those things if we take texts people used one or many languages in the same tweet so there can be bilingual problems, and “hunnnnnnngry” this type of texts we can found in tweets so repeating of same letters also can be a problem when analyzing texts. These problems already arise in previous researches, so when detecting tweets repeating letters and tweets with so many languages is ignored when analyzing texts.

Machine Learning Models for Depression Identification

Among all studies, the use of classifiers to categorize the user as either “depressed” or “not depressed” seems to be a common practice. The following algorithms are mostly used in previous researches in order to identify depression tweets.

Multiple Stepwise Regression algorithm

Multiple Regression

Support Vector Machine

Naive Bayes

Random Forest Hidden Markov Model

References

US Burden of Disease Collaborators, “The state of US health, 1990- 2010: burden of diseases, injuries, and risk factors.,” JAMA, vol. 310, no. 6, pp. 591–608, 2013.

Software Engineer | Tech Enthusiast | Freelance Writer