Public discourse and sentiment during the COVID 19 pandemic: Using Latent Dirichlet Allocation for topic modeling on Twitter.

The study aims to understand Twitter users' discourse and psychological reactions to COVID-19. We use machine learning techniques to analyze about 1.9 million Tweets (written in English) related to coronavirus collected from January 23 to March 7, 2020. A total of salient 11 topics are identified and then categorized into ten themes, including "updates about confirmed cases," "COVID-19 related death," "cases outside China (worldwide)," "COVID-19 outbreak in South Korea," "early signs of the outbreak in New York," "Diamond Princess cruise," "economic impact," "Preventive measures," "authorities," and "supply chain." Results do not reveal treatments and symptoms related messages as prevalent topics on Twitter. Sentiment analysis shows that fear for the unknown nature of the coronavirus is dominant in all topics. Implications and limitations of the study are also discussed.

Clinical determinants of social media use in individuals with schizophrenia.

This study aimed to examine the prevalence of social media use and its association with symptoms in individuals with schizophrenia. 265 individuals with schizophrenia were assessed. Symptoms were assessed on the Positive and Negative Syndrome Scale (PANSS) and the Clinical Assessment Interview for Negative Symptoms (CAINS). Information on social media use was collected. Logistic regressions were used to explore the association between social media use and socio-demographic and clinical characteristics of the participants. Of the 265 study participants, 139 (52.5%) used social media in the last week. Fifty-six (21.1%) of the study participants used more than one social media site in the last week. Facebook was the most popular social media site. Age, highest education level, monthly household income, PANSS negative and depression factor scores were significantly associated with social media use. Amongst negative symptoms, the CAINS motivation-pleasure (MAP) social factor scores were found to be significantly associated with social media use. Our study results suggested that the assessment of social interactions via social media should be considered in the clinical assessment of individuals with schizophrenia. Secondly, our results suggested that the development of treatment programs supported by social media platforms may be useful for certain groups of individuals with schizophrenia. Younger patients with above secondary level education, higher family income and lower symptom severity are likely to be avid users of social media and would be suitable candidates to receive illness related information or clinical interventions via social media.

Impact of social media use on reading levels in third-year student pharmacists.

INTRODUCTION: The majority of Americans report using social media, but there is limited research describing impact of social media on academic performance and reading. Our objectives were to describe the association between social media use and reading levels of third-year student pharmacists (P3), describe the association between reading level and pharmacy school admissions data, and assess texts used in the curriculum for readability. METHODS: This was a prospective, cohort study. Reading level was determined by a standardized test. Social media data were collected via questionnaire. Admissions data were obtained from the admissions office. Readability of texts was assessed using readability software. RESULTS: Eighty-nine student pharmacists completed the study. The average reading level was 16.4. Students reported using social media for an average of 126 min daily. Students reported using an average of four social media sites and spending 88 min weekly on extracurricular reading. Negligible linear correlations were found between reading level and time spent on social media (ρ = 0.063), number of sites used (ρ =0.062), and time spent on extracurricular reading (ρ= 0.130). A moderate correlation (ρ = 0.524) was found between reading level and Pharmacy College Admission Test (PCAT) score. The average readability of guidelines and textbook chapters were 18.1 ±â€¯1.0 and 20.4 ±â€¯0.3, respectively. CONCLUSIONS: In P3 students, reading level was not associated with social media use. However, PCAT scores were positively associated with reading level. Furthermore, the readability of assigned texts exceeded the average reading level of the students.

Tracing medicinal agencies: Antidepressants and life-effects.

This article is inspired by the social life of methods approach, joining a movement among social scientists engaging with 'big data' to contribute to methodological innovation and conceptual development in research and knowledge translation. It explores human-drug associations using a computational tool, Medicine Radar, meanwhile raising questions about the ways a digital device pushes us to rethink how drugs are known in the everyday. Medicine Radar is an apparatus for exploring human-drug associations by means of Suomi24 (Finland24) data, containing 19 million health-related online posts spanning a period of 16 years. Using defined markers, Medicine Radar sorts the medicine talk in health-related discussions, thereby assisting us to 'see' the actions of the drug and human responses to them. This kind of approach distances the drug from the illness experience, drawing attention to the private details of the human-drug relationship. The empirical analysis separates three areas of antidepressant use: articulations of reactions, stabilizing the life effects of drugs and coming to terms with antidepressants. Together, the online posts urge us to think of everyday experience where the effects of drugs - intended or unintended - are always lived. The side effects of antidepressants, including drowsiness, ravenous hunger, loss of sexual desire and emotional numbness, become life effects. As will be demonstrated, the move from conceptualizing such fallout as side effects to understanding them as life effects has political ramifications. The computation tool adds collective weight to antidepressant experiences and calls for politicizing their effects on life.

Patterns of Social Media Use and Their Relationship to Health Risks Among Young Adults.

PURPOSE: Social media use is pervasive among young adults, and different sites have different purposes, features, and audiences. This study identified classes of young adults based on what combination of sites they use and how frequently, and compared their health risk factors and behaviors. METHODS: Latent profile models were developed based on frequency of using 10 sites from a national sample of young adults aged 18-24 years (n = 1,062). Bivariate analyses and multivariable regressions examined the relationship between class membership and alcohol, tobacco, and other drug (ATOD) use, and symptoms of depression and anxiety. RESULTS: The optimal model identified five classes: Low Users (7.9%), High Users (63.1%), Professional Users - high use of LinkedIn (10.1%), Creative Users - high use of Vine and Tumblr (11.5%), and Mainstream Users - high use of Facebook and YouTube (7.4%). Classes differed significantly on ATOD use and depressive symptoms. Compared to High Users, Creative Users had higher odds of using most substances and lower odds of depressive symptoms, Mainstream Users had higher odds of substances used socially (alcohol and hookah), Professional Users had higher odds of using alcohol, cigarettes, and cigars, and Low Users had higher odds of using other drugs (e.g., cocaine and heroin). CONCLUSIONS: A young adult's social media site use profile is associated with ATOD use and depressive symptoms. Use and co-use of certain sites may influence the volume and nature of ATOD-related content and norms young adults experience in social media. Targeting interventions to sites selected based on use patterns associated with each health risk may be effective.

Estimating disease burden using Internet data.

Data on disease burden are often used for assessing population health, evaluating the effectiveness of interventions, formulating health policies, and planning future resource allocation. We investigated whether Internet usage and social media data, specifically the search volume on Google, page view count on Wikipedia, and disease mentioning frequency on Twitter, correlated with the disease burden, measured by prevalence and treatment cost, for 1633 diseases over an 11-year period. We also applied least absolute shrinkage and selection operator to predict the burden of diseases. We found that Google search volume is relatively strongly correlated with the burdens for 39 of 1633 diseases, including viral hepatitis, diabetes mellitus, multiple sclerosis, and hemorrhoids. Wikipedia and Twitter data strongly correlated with the burdens of 15 and 7 diseases, respectively. However, an accurate analysis must consider each condition's characteristics, including acute/chronic nature, severity, familiarity to the public, and the presence of stigma.

Data and systems for medication-related text classification and concept normalization from Twitter: insights from the Social Media Mining for Health (SMM4H)-2017 shared task.

Objective: We executed the Social Media Mining for Health (SMM4H) 2017 shared tasks to enable the community-driven development and large-scale evaluation of automatic text processing methods for the classification and normalization of health-related text from social media. An additional objective was to publicly release manually annotated data. Materials and Methods: We organized 3 independent subtasks: automatic classification of self-reports of 1) adverse drug reactions (ADRs) and 2) medication consumption, from medication-mentioning tweets, and 3) normalization of ADR expressions. Training data consisted of 15 717 annotated tweets for (1), 10 260 for (2), and 6650 ADR phrases and identifiers for (3); and exhibited typical properties of social-media-based health-related texts. Systems were evaluated using 9961, 7513, and 2500 instances for the 3 subtasks, respectively. We evaluated performances of classes of methods and ensembles of system combinations following the shared tasks. Results: Among 55 system runs, the best system scores for the 3 subtasks were 0.435 (ADR class F1-score) for subtask-1, 0.693 (micro-averaged F1-score over two classes) for subtask-2, and 88.5% (accuracy) for subtask-3. Ensembles of system combinations obtained best scores of 0.476, 0.702, and 88.7%, outperforming individual systems. Discussion: Among individual systems, support vector machines and convolutional neural networks showed high performance. Performance gains achieved by ensembles of system combinations suggest that such strategies may be suitable for operational systems relying on difficult text classification tasks (eg, subtask-1). Conclusions: Data imbalance and lack of context remain challenges for natural language processing of social media text. Annotated data from the shared task have been made available as reference standards for future studies (http://dx.doi.org/10.17632/rxwfb3tysd.1).

Characterizing JUUL-related posts on Twitter.

BACKGROUND: As vaping rapidly becomes more prevalent, social media data can be harnessed to capture individuals' discussions of e-cigarette products quickly. The JUUL vaporizer is the latest advancement in e-cigarette technology, which delivers nicotine to the user from a device that is the size and shape of a thumb drive. Despite JUUL's growing popularity, little research has been conducted on JUUL. Here we utilized Twitter data to determine the public's early experiences with JUUL describing topics of posts. METHODS: Twitter posts containing the term "JUUL" were obtained for 1 April 2107 to 14 December 2017. Text classifiers were used to identify topics in posts (n = 81,689). RESULTS: The most prevalent topic wasPerson Tagging (use of @username to tag someone in a post) at 20.48% followed by Pods (mentions of JUUL's refill cartridge) at 14.72% and Buying (mentions of purchases) at 10.49%. The topic School (posts indicative of using JUUL or seeing someone use JUUL while at elementary, middle, or high school) comprised 3.66% of posts. The topic of Quit Smoking was rare at 0.29%. CONCLUSIONS: Data from social media may be used to extend the surveillance of newly introduced vaping products. Findings suggest Twitter users are bonding around, and inquiring about, JUUL on social media. JUUL's discreetness may facilitate its use in places where vaping is prohibited. Educators may be in need of training on how to identify JUUL in the classroom. Despite JUUL's branding as a smoking alternative, very few Twitter users mentioned smoking cessation with JUUL.

Detecting depression stigma on social media: A linguistic analysis.

BACKGROUND: Efficient detection of depression stigma in mass media is important for designing effective stigma reduction strategies. Using linguistic analysis methods, this paper aims to build computational models for detecting stigma expressions in Chinese social media posts (Sina Weibo). METHODS: A total of 15,879 Weibo posts with keywords were collected and analyzed. First, a content analysis was conducted on all 15,879 posts to determine whether each of them reflected depression stigma or not. Second, using four algorithms (Simple Logistic Regression, Multilayer Perceptron Neural Networks, Support Vector Machine, and Random Forest), two groups of classification models were built based on selected linguistic features; one for differentiating between posts with and without depression stigma, and one for differentiating among posts with three specific types of depression stigma. RESULTS: First, 967 of 15,879 posts (6.09%) indicated depression stigma. 39.30%, 15.82%, and 14.99% of them endorsed the stigmatizing view that "People with depression are unpredictable", "Depression is a sign of personal weakness", and "Depression is not a real medical illness", respectively. Second, the highest F-Measure value for differentiating between stigma and non-stigma reached 75.2%. The highest F-Measure value for differentiating among three specific types of stigma reached 86.2%. LIMITATIONS: Due to the limited and imbalanced dataset of Chinese Weibo posts, the findings of this study might have limited generalizability. CONCLUSIONS: This paper confirms that incorporating linguistic analysis methods into online detection of stigma can be beneficial to improve the performance of stigma reduction programs.

Ontology-based automatic identification of public health-related Turkish tweets.

Social media analysis, such as the analysis of tweets, is a promising research topic for tracking public health concerns including epidemics. In this paper, we present an ontology-based approach to automatically identify public health-related Turkish tweets. The system is based on a public health ontology that we have constructed through a semi-automated procedure. The ontology concepts are expanded through a linguistically motivated relaxation scheme as the last stage of ontology development, before being integrated into our system to increase its coverage. The ultimate lexical resource which includes the terms corresponding to the ontology concepts is used to filter the Twitter stream so that a plausible tweet subset, including mostly public-health related tweets, can be obtained. Experiments are carried out on two million genuine tweets and promising precision rates are obtained. Also implemented within the course of the current study is a Web-based interface, to track the results of this identification system, to be used by the related public health staff. Hence, the current social media analysis study has both technical and practical contributions to the significant domain of public health.

Identifying Twitter influencer profiles for health promotion in Saudi Arabia.

New media platforms, such as Twitter, provide the ideal opportunity to positively influence the health of large audiences. Saudi Arabia has one of the highest number of Twitter users of any country, some of whom are very influential in setting agendas and contributing to the dissemination of ideas. Those opinion leaders, both individuals and organizations, influential in the new media environment have the potential to raise awareness of health issues, advocate for health and potentially instigate change at a social level. To realize the potential of the new media platforms for public health, the function of opinion leaders is key. This study aims to identify and profile the most influential Twitter accounts in Saudi Arabia. Multiple measures, including: number of followers and four influence scores, were used to evaluate Twitter accounts. The data were then filtered and analysed using ratio and percentage calculations to identify the most influential users. In total, 99 Saudi Twitter accounts were classified, resulting in the identification of 25 religious men/women, 16 traditional media, 14 sports related, 10 new media, 6 political, 6 company and 4 health accounts. The methods used to identify the key influential Saudi accounts can be applied to inform profile development of Twitter users in other countries.

Two Studies on Twitter Networks and Tweet Content in Relation to Amyotrophic Lateral Sclerosis (ALS): Conversation, Information, and 'Diary of a Daily Life'.

To date, there is no research examining how adults with Amyotrophic Lateral Sclerosis (ALS) or Motor Neurone Disease (MND) and severe communication disability use Twitter, nor the use of Twitter in relation to ALS/MND beyond its use for fundraising and raising awareness. In this paper we (a) outline a rationale for the use of Twitter as a method of communication and information exchange for adults with ALS/MND, (b) detail multiple qualitative and quantitative methods used to analyse Twitter networks and tweet content in the our studies, and (c) present the results of two studies designed to provide insights on the use of Twitter by an adult with ALS/MND and by #ALS and #MND hashtag communities in Twitter. We will also discuss findings across the studies, implications for health service providers in Twitter, and directions for future Twitter research in relation to ALS/MND.

Tweeting the Meeting: Twitter Use at The American Society of Breast Surgeons Annual Meeting 2013-2016.

BACKGROUND: Twitter social media is being used to disseminate medical meeting information. Meeting attendees and other interested parties have the ability to follow and participate in conversations related to meeting content. We analyzed Twitter activity generated from the 2013-2016 American Society of Breast Surgeons Annual Meetings. METHODS: The Symplur Signals database was used to determine number of tweets, tweets per user, and impressions for each meeting. The number of unique physicians, patients/caregivers/advocates, and industry participants was determined. Physician tweeters were cross-referenced with membership and attendance rosters. Tweet transcripts were analyzed for content and tweets were categorized as either scientific, social, administrative, industry promotion, or irrelevant. RESULTS: From 2013 to 2016, the number of tweets increased by 600 %, the number of Twitter users increased by 450 %, and the number of physician tweeters increased by 457 %. The number of impressions (tweets × followers) increased from more than 3.5 million to almost 20.5 million, an increase of 469 %. The majority of tweets were informative (70-80 %); social tweets ranged from 13 to 23 %. A small percentage (3-6 %) of tweets were related to administrative matters. There were very few industry or irrelevant tweets. CONCLUSIONS: Twitter social media use at the American Society of Breast Surgeons annual meeting showed a substantial increase during the time period evaluated. The use of Twitter during professional meetings is a tremendous opportunity to share information. The authors feel that medical conference organizers should encourage Twitter participation and should be educating attendees on the proper use of Twitter.

A content analysis of tweets about high-potency marijuana.

INTRODUCTION: "Dabbing" involves heating extremely concentrated forms of marijuana to high temperatures and inhaling the resulting vapor. We studied themes describing the consequences of using highly concentrated marijuana by examining the dabbing-related content on Twitter. METHODS: Tweets containing dabbing-related keywords were collected from 1/1-1/31/2015 (n=206,854). A random sample of 5000 tweets was coded for content according to pre-determined categories about dabbing-related behaviors and effects experienced using a crowdsourcing service. An examination of tweets from the full sample about respiratory effects and passing out was then conducted by selecting tweets with relevant keywords. RESULTS: Among the 5000 randomly sampled tweets, 3540 (71%) were related to dabbing marijuana concentrates. The most common themes included mentioning current use of concentrates (n=849; 24%), the intense high and/or extreme effects from dabbing (n=763; 22%) and excessive/heavy dabbing (n=517; 15%). Extreme effects included both physiological (n=124/333; 37%) and psychological effects (n=55/333; 17%). The most common physiologic effects, passing out (n=46/333; 14%) and respiratory effects (n=30/333; 9%), were then further studied in the full sample of tweets. Coughing was the most common respiratory effect mentioned (n=807/1179; 68%), and tweeters commonly expressed dabbing with intentions to pass out (416/915; 45%). CONCLUSIONS: This study adds to the limited understanding of marijuana concentrates and highlights self-reported physical and psychological effects from this type of marijuana use. Future research should further examine these effects and the potential severity of health consequences associated with concentrates.

A Framework for Classifying Online Mental Health-Related Communities With an Interest in Depression.

Mental illness has a deep impact on individuals, families, and by extension, society as a whole. Social networks allow individuals with mental disorders to communicate with others sufferers via online communities, providing an invaluable resource for studies on textual signs of psychological health problems. Mental disorders often occur in combinations, e.g., a patient with an anxiety disorder may also develop depression. This co-occurring mental health condition provides the focus for our work on classifying online communities with an interest in depression. For this, we have crawled a large body of 620 000 posts made by 80 000 users in 247 online communities. We have extracted the topics and psycholinguistic features expressed in the posts, using these as inputs to our model. Following a machine learning technique, we have formulated a joint modeling framework in order to classify mental health-related co-occurring online communities from these features. Finally, we performed empirical validation of the model on the crawled dataset where our model outperforms recent state-of-the-art baselines.

Development of an Adolescent Depression Ontology for Analyzing Social Data.

Depression in adolescence is associated with significant suicidality. Therefore, it is important to detect the risk for depression and provide timely care to adolescents. This study aims to develop an ontology for collecting and analyzing social media data about adolescent depression. This ontology was developed using the 'ontology development 101'. The important terms were extracted from several clinical practice guidelines and postings on Social Network Service. We extracted 777 terms, which were categorized into 'risk factors', 'sign and symptoms', 'screening', 'diagnosis', 'treatment', and 'prevention'. An ontology developed in this study can be used as a framework to understand adolescent depression using unstructured data from social media.

E-Patient Reputation in Health Forums.

Online health forums are increasingly used by patients to get information and help related to their health. However, information reliability in these forums is unfortunately not always guaranteed. Obviously, consequences of self-diagnosis may be severe on the patient's health if measures are taken without consulting a doctor. Many works on trust issues related to social media have been proposed, but most of them mainly focus only on the structure part of the social network (number of posts, number of likes, etc.). In the case of online health forums, a lot of trust and distrust is expressed inside the posted messages and cannot be inferred by only considering the structure. In this study, we rather suggest inferring the user's trustworthiness from the replies he receives in the forum. The proposed method is divided into three main steps: First, the recipient(s) of each post must be identified. Next, the trust or distrust expressed in these posts is evaluated. Finally, the user's reputation is computed by aggregating all the posts he received. Conducted experiments using a manually annotated corpus are encouraging.

Automatically Expanding the Synonym Set of SNOMED CT using Wikipedia.

Clinical terminologies and ontologies are often used in natural language processing/understanding tasks as a method for semantically tagging text. One ontology commonly used for this task is SNOMED CT. Natural language is rich and varied: many different combinations of words may be used to express the same idea. It is therefore essential that ontologies and terminologies have a rich set of synonyms. One source of synonyms is Wikipedia. We examine methods for aligning concepts in SNOMED CT with articles in Wikipedia so that newly-found synonyms may be added to SNOMED CT. Our experiments show promising results and provide guidance to researchers who wish to use Wikipedia for similar tasks.

Identifying Diseases, Drugs, and Symptoms in Twitter.

Social media sites, such as Twitter, are a rich source of many kinds of information, including health-related information. Accurate detection of entities such as diseases, drugs, and symptoms could be used for biosurveillance (e.g. monitoring of flu) and identification of adverse drug events. However, a critical assessment of performance of current text mining technology on Twitter has not been done yet in the medical domain. Here, we study the development of a Twitter data set annotated with relevant medical entities which we have publicly released. The manual annotation results show that it is possible to perform high-quality annotation despite of the complexity of medical terminology and the lack of context in a tweet. Furthermore, we have evaluated the capability of state-of-the-art approaches to reproduce the annotations in the data set. The best methods achieve F-scores of 55-66%. The data analysis and the preliminary results provide valuable insights on identifying medical entities in Twitter for various applications.