International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 06 Issue: 03 | Mar 2019 www.irjet.net p-ISSN: 2395-0072 © 2019, IRJET | Impact Factor value: 7.211 | ISO 9001:2008 Certified Journal | Page 2934 INTELLIGENT DEPRESSION DETECTION SYSTEM B.Yamini 1 , S.Aruljothi 2 , S.M Madhumitha 3 1 Assistant Professor, Computer Science Department, JEPPIAAR SRR Engineering College, Tamil Nadu 2,3 UG Student, Computer Science Department, JEPPIAAR SRR Engineering College, Tamil Nadu ----------------------------------------------------------------------***--------------------------------------------------------------------- Abstract-- Depression is becoming a serious leading mental health problem worldwide. It is a cause of psychological disorders and economic burden to a country. Even though it is a serious psychological problem, less than a half of those who have this emotional problem gained access to mental health service. This could be a result of many factors including having lack awareness about the disease. New solutions are needed to tackle this issue. We propose a system is to develop prediction models to classify users according to depression levels from Facebook data by employing Natural Language Processing (NLP) techniques where people use Facebook as a tool for sharing opinions, feelings and life events. Keywords -- Natural Language Processing, Statistical analysis, Facebook site, Depression screening, User Generated Content (UGC). 1. Introduction Social media can be exploited due to the large amount of information, which refers to user behavioral attributes. Getting use of that information to predict the social media users’ mental health level can help psychiatrist, family or friends to get the right medical advice and therapy on time to the depressed user. WHO ranks the depression as one of the most devastating diseases in the world. In addition, about two thirds of depressed people do not seek appropriate treatments, which lead to major consequences. The medical science relies on asking the patients questions about their circumstances, which does not diagnose the depression in a precise way. The patient has to attend more than one session during a period of three weeks. The classification of a not depressed condition as a depressed is a False Positive problem. However, researchers found that the Electronic Health Record (EHR) systems are not optimally designed to handle integrating behavioral health and primary care. EHRs lack to support documenting and tracking data for behavioural health conditions such as depression. Most of the people use social media to express their feelings, emotions. Many researchers have been successfully proving that social media has been successfully used to maintain people's mental health. By mining the social media posts of users, we may get a complete image of the user natural behaviour, thinking style, interactions, guilt feeling, worthlessness, loneliness, and helplessness. Retrieving such behavioural attributes, show symptoms of depression on the social media users, which could be used to predict if the user is depressed or not. Psychiatrist, parents, and friends, could track the user depression by the proposed tool, which will save the time before the depressed user could get into major depression phase. 2. Related Work Several approaches have been studied for collecting social media data with associated information about the users’ mental health. Participants are either recruited to take a depression survey and share their Facebook or Twitter data, or data is collected from existing public online sources. These sources include searching public Tweets for keywords to identify (and obtain all Tweets from) users who have shared their mental health diagnosis, user language on mental illness related forums, or through collecting public Tweets that mention mental illness keywords for annotation. The approaches using public data have the advantage that much larger samples can, in principle, be collected faster and more cheaply than through the administration of surveys, though survey-based assessment generally provides a higher degree of validity. We first compare studies that attempt to distinguish mentally ill users from neurotypical controls. A. Prediction Based on Survey Responses Psychometric self-report surveys for mental illness have a high degree of validity and reliability. In psychological and epidemiological research, self- report surveys are second only to clinical interviews, which no social media study to date has used as an outcome measure. We discuss five studies that predict survey assessed depression status by collecting participants’ responses to depression surveys in conjunction with their social media data. The most cited study used Twitter activity to examine network and language data preceding a recent episode of depression. The presence of depression was established through participants reporting the occurrence and recent date of a depressive episode, combined with scores on the Center forEpidemiologic Studies Depression Scale Revised (CES-D) and Beck’s Depression Inventory (BDI). This study revealed several distinctions in posting activity by depressed users, including: diurnal cycles, more negative emotion, less social interaction, more self- focus, and mentioning depression-related terms throughout the year preceding depression onset. Reece et al. predicted user depression and posttraumatic stress-disorder (PTSD) status from text and Twitter meta-data that preceded a reported first