To execute the proposed project, First, we gather raw data set of depressive and anxious tweets. Articulating the problem, defining data required, collecting and combining data from different sources are the 3 main steps under data collecting. Also, we are planning to discuss with medical officers and other experts in psychiatry and obtain local data-sets if possible. Then formatting, cleaning, and sampling will be done in data set preprocessing. We will research and use suitable Natural Language Processing (NLP) technique (ex: Neural Networks/Deep Learning) to determine the structural role of words in the sentence and categorize them accordingly using the preprocessed data. We will study and use knowledge of the researches conducted on NLP classifying (ex: From Sentiment Analysis to Emotion Recognition NLP [7]) to improve our system.

Then the depression measuring module will be developed with the help of machine learning. We plan to research and work with the most suitable classifier (ex: SVM classifier, Naïve Bayes classifier) for our project.

We are planning to collect the test data set from tweets of individuals by using Twitter API or an alternative method such as a portable framework for testing web applications (ex: Selenium)  and use the collected new test data on our module, test our module and use the test results for the betterment of the people who suffer from this major health problem of depression. 

Figure shows the interaction among those modules.

.

Though depression is considered a major health problem, enough steps haven’t been taken to reduce its consequences, like reducing the number of people who commit suicide every year. Depression plays a role in more than one-half of all suicide attempts, whereas the lifetime risk of suicide among untreated depressive disorder patients is nearly 20%[1]. Many research have found that nearly 4.4% of the global population is suffering from depression. When considering gender, females found more common with depression (5.1%) while males are less (3.6%). The estimated total cases in Sri Lanka are 802,321 and 4.1% of the total population.[2]
A survey conducted from June to August 2020 among 130 countries across WHO’s six regions evaluates how the provision of mental, neurological, and substance use services has changed due to COVID-19, the types of services that have been disrupted, and how countries are adapting to overcome these challenges.[3] WHO has issued guidance to countries on maintaining essential services including mental health services during COVID 19 and it is recommended by WHO to have special consideration and allocate more resources for mental health. During surveys, 89% of the countries have reported that mental health and psychosocial support is among their national COVID-19 response plans. But it is found that only 17% of these countries have funded fully or additionally in order to cover these activities.  It is not enough to allocate an amount of 2% from a national health budget for mental health. As mental health still gets less than 1% of international aid earmarked for health, international funders also need to contribute more to uplift mental health throughout the world.
And it is essential to identify people with depression. There are medical researches on Text-Based Detection and measuring of Depression. In our research, we plan to use the help of such researches (ex: Text-Based Detection of the Risk of Depression [4]), develop an AI-based system, and use it to analyze Twitter social media tweets to measure individuals' depression levels.

[1] I.Gotlib and C.Hammen, “Handbook of Depression,” Guilford Press, New York, 2002.
[2] The Epidemiology Unit, Ministry of Health, Nutrition & Indigenous Medicine, “Weekly Epidemiological Report”, Vol 44 No 15, 2017 April 8th- 14th.                                          [Online].Available:  http://www.epid.gov.lk/web/images/pdf/wer/2017/vol_44_no_15-english.pdf  ,[Accessed 24 October 2020].
[3] “COVID-19 disrupting mental health services in most countries, WHO survey,” World Health Organization. [Online]. Available: https://www.who.int/news/item/05-10-2020-covid-19-disrupting-mental-health-services-in-most-countries-who-survey.  ,[Accessed 25 October 2020].
[4] Jana M.Havigerova, Jiri Haviger, Dalibor Kucera and Petra Hoffmannova, “Text-Based Detection of the Risk of Depression,” March 18, 2019. [Online]. Available: https://www.frontiersin.org/articles/10.3389/fpsyg.2019.00513/full  ,[Accessed 27 October 2020].