142nd APHA Annual Meeting and Exposition

Annual Meeting Recordings are now available for purchase

Social media and big data: Can Tweet moods predict illness and hospital/doc visits in a region?

142nd APHA Annual Meeting and Exposition (November 15 - November 19, 2014): http://www.apha.org/events-and-meetings/annual
Tuesday, November 18, 2014

Priya Nambisan, PhD , Department of Health Informatics and Administration, University of Wisconsin - Milwaukee, Milwaukee, WI
Zhihui Luo, PhD , Department of Health Informatics and Administration, University of Wisconsin - Milwaukee, Milwaukee, WI
Akshat Kapoor , Health Informatics and Administration, University of Wisconsin - Milwaukee, Milwaukee, WI
Background:Depression has been associated with increased morbidity and mortality from illnesses such as heart disease and diabetes. It is also associated with adverse health habits such as smoking, over-eating, sedentary lifestyle, which in turn lead to obesity and other health issues. Thus, it is quite evident from prior studies that mood disorders can predict physical illnesses to some extent. There have been some studies that have tried to glean moods and even depression from tweets; however, these studies did not connect them to disease or illness or to hospital/doc visits. 

Objective: The objective of this research is to analyze tweets regarding depression – characterized by sadness, feelings of loneliness, gloominess and other characteristics associated with depression, such as lack of concentration, suicidal thoughts, dejection etc., to see if those tweets predict illnesses and consequently lead to hospital/doc visits. 

Method: We developed a semantic lexicon to capture user vocabularies and phrases that can characterize depression and moods. We draw from sentiment analysis literature and also use vocabulary that is ‘trending’ in tweets regarding moods. Next, we evaluate approximately 200 million tweets and use a software program to assign a depression score – ‘highly depressed’ to ‘not at all depressed’. Finally, statistical association analysis (e.g., Apriori) will be used to connect the depression tweets with disease/ illness tweets or doc/hospital visit tweets. 

Results: The study findings are expected to hold important implications for how we can use big data from social media to predict disease outcomes and healthcare service utilization. This can be used to predict whether there are regional or seasonal differences in moods and whether those differences impact disease outcomes. This study is the first of its kind and holds potential to unveil critical social and clinical factors related to depression, which in turn has considerable implications for public health.

Learning Areas:

Communication and informatics
Planning of health education strategies, interventions, and programs
Provision of health care to the public
Social and behavioral sciences

Learning Objectives:
Identify the specific outcomes of using Tweet moods to predict disease. List the key factors that would determine the overall success of using Tweets to predict public health trends. Discuss the key challenges and issues when using big data from social media as a public health tool.

Keyword(s): Social Media, Depression

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I am a doctoral student working in the area of health sciences and have a background in public health. I have also collaborated with various faculty and researchers over various projects in the field of public health, and have even co-authored publications.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.