Table 2.

Study characteristics

Author<	Purpose	Data Source	Text Type	Number of Docum ents	Number of Users
Brennan & Aron	To evaluate the application of MetaMap for detecting the presence of terms found in the UMLS within the electronic messages of patients	Internet-based home care post-discharge support service intervention	Emails sent by patients to a clinical nurse	241 electronic messages	Not reported
Portier et al, 201333	To examine whether sentiment change is influenced by the main topic of the initiating post	Online peer support community for cancer patients (ACS’s Cancer Survivors Network) forums for breast and colorectal cancer	Discussion posts	29,384 threaded discussions	Not reported
Freifeld et al, 201427	To evaluate the level of concordance between Twitter posts mentioning adverse event-like reactions and spontaneous reports received by a regulatory agency	Twitter	Tweets with mentions of 23 drugs and 4 vaccines and resemblance to adverse events	4,401 tweets	Not reported
Gupta et al, 201419	To extract symptoms and conditions as well as drugs and treatments from patient-authored text by learning lexico-syntactic patterns from data annotated with seed dictionaries	Online community forum (MedHelp.org) for asthma, ENT, adult type II diabetes, acne, and breast cancer	Sentences	680,071 sentences	Not reported
Park & Ryu, 201425	To evaluate the possibility of using text-mining to identify clinical distinctions and patient concerns in online memoirs posted by patients with fibromyalgia	Online social networking community (Experienceproject.com) forum with title “I Have Fibromyalgi a”	Patient narratives	399 narratives	Not reported
Janies et al, 201528	To create a web-based workflow application that uses chief complaints from Twitter as a syndromic surveillance tool and correlates outbreak signals to pathogens known to circulate a geographic area	Twitter	Tweets	>1,000,000 tweets	Not reported
Jimeno-Yepes et al, 201520	To develop an annotated data set from Twitter feeds that can be used to train and evaluate methods to recognize mentions of diseases, symptoms, and pharmacologic substances in social media	Twitter	Tweets with 2 out of 3 entity types – diseases symptoms, or pharma cological substances	1,300 tweets	Not reported
Karmen et al, 201521	To develop a method that detects symptoms of depression in free text from social media sources	Online public mental health message board (Psycho-Babble) “Grief” forum	User posts within a 20–200 word interval	1,304 posts	Not reported
Liu & Chen, 201523	To develop a research framework for patient reported adverse drug event extraction	3 online patient forums for diabetes (ADA online community, Diabetes Forum, and Diabetes Forums) and 1 for heart disease (MedHelp.org)	Patient discussi on posts	1,072,474 posts	Not reported
Nikfarjam et al, 201524	To design a machine learning-based approach to extract mentions of adverse drug reactions from social media text	Twitter and a health related social network (DailyStrength)	Tweets and user posts for 81 widely used drugs	1,784 tweets and 6,279 post from DailyStrength	Not reported
Tighe et al, 201532	To examine the type, context, and dissemination of pain-related tweets from 50 cities around the world	Twitter	Tweets with mention s of pain	47,958 tweets	Not reported
Eshleman & Singh, 201618	To described a framework based on graph-theoretic modeling of drug-effect relationships drawn from various data sources	Twitter	Tweets with mentions of 200 commonly prescribed drugs	157,735 tweets	7,981
Lee & Donovan, 201635	To understand the symptom experiences and strategies that were associated with fatigue management among women with ovarian cancer	Web-based ovarian cancer symptom management intervention	Patient respons es to prompts	Not reported	165
Marshall et al, 201629	To compare symptom cluster patterns derived from messages on a breast cancer forum with those from a symptom checklist completed by breast cancer survivors participating in a research study	Online community forum (MedHelp.or g) for breast cancer	Messag e posts	50,426 posts	12,991
Topaz et al, 201630	To compare electronic health record data and social media data to clinician-reported adverse drug reactions and patients’ concerns regarding aspirin and atorvastatin	Treato Ltd (treato.com) database of health-related websites, forums, blogs, and Treato discussion platform	List of potential adverse drug reactions for aspirin and atorvast atin	42,594 potential adverse drug reactions	Not reported
Sunkureddi et al, 201634	To describe patient experiences reported online to better understand the day-to-day disease burden of ankylosing spondylitis	52 online sources, including social networking sites, patient- physician Q&A sites, and ankylosing spondylitis forums	Patient narratives	34,780 narratives	3,449
Cocos et al, 201717	To develop a scalable, deep-learning approach for adverse drug reaction detection in social media data	Twitter	Tweets for 81 widely used drugs and 44 ADHD drugs	844 tweets	Not reported
Cronin et al, 201736	To develop automated patient portal message classifiers for communication type (i.e., informational, logistic, social, medical, and other)	Patient portal (My Health at Vanderbilt)	Portal messages	3,253 messages	3,116
Lamy et al, 201722	To examine synthetic cannabinoid receptor agonist-related effects and their variations through a longitudinal content analysis of web-forum data	3 drug focused web forums (Bluelight.or g and 2 anonymized forums)	User posts related to synthetic cannabinoid receptor agonists	19,052 posts	2,543
Lu et al, 201731	To develop a content analysis method for stakeholder analysis, topic analysis, and sentiment analysis health care social media	Online community forum (MedHelp.org) for lung cancer, diabetes, and breast cancer	Messag e posts	138,161 posts	39,606
Patel et al, 201726	To detect and quantify glucocorticoid-related adverse events using a computerized system for automated suspected adverse drug reaction detection from narrative text in Twitter	Twitter	Tweets with mentions of	15,730	Not reported
			prednis one or prednis olone	tweets

Note.

^<Studies have been arranged in chronological order to assess trends over time; ACS=American Cancer Society; ADA=American Diabetes Association; ADHD=attention deficit hyperactivity disorder; ENT=ear, nose, and throat; UMLS=Unified Medical Language System