Using social media to monitor mental health discussions - evidence from Twitter

Exploring temporal suicidal behavior patterns on social media: Insight from Twitter analytics

A valid mechanism for suicide detection and intervention to a wider population online has not yet been fully established. With the increasing suicide rate, we proposed an approach that aims to examine temporal patterns of potential suicidal ideations and behaviors on Twitter to better understand their risk factors and time-varying features. It identifies latent suicide topics and then models the suicidal topic-related score time series to quantitatively represent behavior patterns on Twitter. After evaluated on a collection of suicide-related tweets in 2016, 13 key risk factors were discovered and the temporal patterns of suicide behavior on different days during 1 week were identified to highlight the distinct time-varying features related to different risk factors. This study is practical to help public health services and others to develop refined prevention strategies, to monitor and support a population of high-risk at right moments.

Celebrity suicide on Twitter: Activity, content and network analysis related to the death of Swedish DJ Tim Bergling alias Avicii

BACKGROUND

Media recommendations for suicide reporting are recommended to prevent imitative suicide but little is known about social media reactions to different revelations about celebrity suicide.

METHODS

Using the Twitter Application Programming Interface (API), we recorded public tweets mentioning Avicii from the day when his death was reported (N = 2,865,292). We compared that data with a dataset of random tweets. Furthermore, we recorded tweets including suicide in 124 languages before Avicii's death (N = 5,939,107). We processed English tweets mentioning Avicii with the Linguistic Inquiry and Word Count (LIWC) to quantify the frequency of affects and related linguistic signals. We also processed the text of tweets to detect tweets mentioning the suicide method, and we retrieved the list of followers of users who tweeted about the method. We constructed reply networks from the dataset, analysing three networks corresponding to the major news events about Avicii's death.

RESULTS

Avicii's suicide sparked immediate strong interest with both positive (χ² = 781.06, p < 10^-6) and negative emotional expressions (χ² = 1518.5, p < 10^-6) in comparison to baseline levels. Subsequent revelations were associated with smaller peaks with mainly negative emotional content after Avicii's death was revealed as a suicide (χ² = 33.2, p < 10^-6 and after news about the suicide method (χ² = 274.93, p < 10^-6). Tweeting about the suicide method was infrequent, but twitter users who covered the method had more followers that users who did not (D = 0.1675, p < 10^-6; t = 19.87, p < 10^-6), and a noteworthy number of users had considerable exposure to the suicide method.

LIMITATIONS

This was a descriptive analysis.

CONCLUSIONS

Twitter users showed strong interest in news about Avicii's death and Avicii's suicide, but less so in the suicide method, and showed distinct tweeting behaviours based on the different revelations.

Crisis Text Line use following the release of Netflix series 13 Reasons Why Season 1: Time-series analysis of help-seeking behavior in youth

The availability of near real-time data from Crisis Text Line (CTL) and other technology-based platforms on crisis events provides an opportunity for targeted interventions prior to serious mental health outcomes (e.g., suicide, self-harm). This study examined the association between the release of the popular Netflix series 13 Reasons Why (13RW) and CTL usage in a national sample of youth in the US. We implemented interrupted time-series, autoregressive integrated moving average (ARIMA) modeling to examine this association at a daily scale. We observed a significant but momentary rise in CTL conversation volume following the release of 13RW on April 5 and 6, 2017 followed by a significant reduction (12.7%) in conversation volume for the overall study period. This reduction in call volume was sustained for 49 days and is the most sustained reduction in conversation volume in the 365 day dataset. This unexpected trough in conversation volume is concerning in light of elevated search engine volume for terms indicating an increase in suicidal thoughts in the days following the release of the show (Ayers et al., 2017). CTL was featured by the show as a resource for viewers in the recently released Season 2, and our results highlight the reasoning and need for such promotion. Future work should explore whether the promotion of CTL in Season 2 positively impacted conversation volume, as there is a clear need to harness the power of these digital technologies to detect population-based trends in mental health and expand the reach of life saving services.

Period of Measurement in Time-Series Predictions of Disease Counts from 2007 to 2017 in Northern Nevada: Analytics Experiment

BACKGROUND

The literature in statistics presents methods by which autocorrelation can identify the best period of measurement to improve the performance of a time-series prediction. The period of measurement plays an important role in improving the performance of disease-count predictions. However, from the operational perspective in public health surveillance, there is a limitation to the length of the measurement period that can offer meaningful and valuable predictions.

OBJECTIVE

This study aimed to establish a method that identifies the shortest period of measurement without significantly decreasing the prediction performance for time-series analysis of disease counts.

METHODS

The data used in this evaluation include disease counts from 2007 to 2017 in northern Nevada. The disease counts for chlamydia, salmonella, respiratory syncytial virus, gonorrhea, viral meningitis, and influenza A were predicted.

RESULTS

Our results showed that autocorrelation could not guarantee the best performance for prediction of disease counts. However, the proposed method with the change-point analysis suggests a period of measurement that is operationally acceptable and performance that is not significantly different from the best prediction.

CONCLUSIONS

The use of change-point analysis with autocorrelation provides the best and most practical period of measurement.

Text Mining Mental Health Reports for Issues Impacting Today's College Students: Qualitative Study

BACKGROUND

A growing number of college students are experiencing personal circumstances or encountering situations that feel overwhelming and negatively affect their academic studies and other aspects of life on campus. To meet this growing demand for counseling services, US colleges and universities are offering a growing variety of mental health services that provide support and services to students in distress.

OBJECTIVE

In this study, we explore mental health issues impacting college students using a corpus of news articles, foundation reports, and media stories. Mental health concerns within this population have been on the rise. Uncovering the most salient themes articulated in current news and literature reports can better enable higher education institutions to provide health services to its students.

METHODS

We used SAS Text Miner to analyze 165 references that were published from 2010 to 2015 and focused on mental health among college students. Key clusters were identified to reveal the themes that were most significant to the topic.

RESULTS

The final cluster analysis yielded six themes in students' mental health experiences in higher education (ie, age, race, crime, student services, aftermath, victim). Two themes, increasing demand for student services provided by campus counseling centers (113/165, 68.5%) and the increased mental health risks faced by racial and ethnic minorities (30/165, 18.2%), dominated the discourse.

CONCLUSIONS

Higher education institutions are actively engaged in extending mental health services and offering targeted outreach to students of color. Cluster analysis identified that institutions are devoting more and innovative resources in response to the growing number students who experience mental health concerns. However, there is a need to focus on proactive approaches to mitigate the causes of mental health and the aftermath of a negative experience, particularly violence and sexual assault. Such strategies can potentially influence how students navigate their health information seeking and how information and communication technologies, including mobile apps, can partially address the needs of college students.

Toward Real-Time Infoveillance of Twitter Health Messages

There is growing interest in conducting public health research using data from social media. In particular, Twitter "infoveillance" has demonstrated utility across health contexts. However, rigorous and reproducible methodologies for using Twitter data in public health are not yet well articulated, particularly those related to content analysis, which is a highly popular approach. In 2014, we gathered an interdisciplinary team of health science researchers, computer scientists, and methodologists to begin implementing an open-source framework for real-time infoveillance of Twitter health messages (RITHM). Through this process, we documented common challenges and novel solutions to inform future work in real-time Twitter data collection and subsequent human coding. The RITHM framework allows researchers and practitioners to use well-planned and reproducible processes in retrieving, storing, filtering, subsampling, and formatting data for health topics of interest. Further considerations for human coding of Twitter data include coder selection and training, data representation, codebook development and refinement, and monitoring coding accuracy and productivity. We illustrate methodological considerations through practical examples from formative work related to hookah tobacco smoking, and we reference essential methods literature related to understanding and using Twitter data.

Youth Mental Health Services Utilization Rates After a Large-Scale Social Media Campaign: Population-Based Interrupted Time-Series Analysis

BACKGROUND

Despite the uptake of mass media campaigns, their overall impact remains unclear. Since 2011, a Canadian telecommunications company has operated an annual, large-scale mental health advocacy campaign (Bell Let's Talk) focused on mental health awareness and stigma reduction. In February 2012, the campaign began to explicitly leverage the social media platform Twitter and incented participation from the public by promising donations of Can $0.05 for each interaction with a campaign-specific username (@Bell_LetsTalk).

OBJECTIVE

The intent of the study was to examine the impact of this 2012 campaign on youth outpatient mental health services in the province of Ontario, Canada.

METHODS

Monthly outpatient mental health visits (primary health care and psychiatric services) were obtained for the Ontario youth aged 10 to 24 years (approximately 5.66 million visits) from January 1, 2006 to December 31, 2015. Interrupted time series, autoregressive integrated moving average modeling was implemented to evaluate the impact of the campaign on rates of monthly outpatient mental health visits. A lagged intervention date of April 1, 2012 was selected to account for the delay required for a patient to schedule and attend a mental health-related physician visit.

RESULTS

The inclusion of Twitter into the 2012 Bell Let's Talk campaign was temporally associated with an increase in outpatient mental health utilization for both males and females. Within primary health care environments, female adolescents aged 10 to 17 years experienced a monthly increase in the mental health visit rate from 10.2/1000 in April 2006 to 14.1/1000 in April 2015 (slope change of 0.094 following campaign, P<.001), whereas males of the same age cohort experienced a monthly increase from 9.7/1000 to 9.8/1000 (slope change of 0.052 following campaign, P<.001). Outpatient psychiatric services visit rates also increased for both male and female adolescents aged 10 to 17 years post campaign (slope change of 0.005, P=.02; slope change of 0.003, P=.005, respectively). For young adults aged 18 to 24 years, females who used primary health care experienced the most significant increases in mental health visit rates from 26.5/1000 in April 2006 to 29.2/1000 in April 2015 (slope change of 0.17 following campaign, P<.001).

CONCLUSIONS

The 2012 Bell Let's Talk campaign was temporally associated with an increase in the rate of mental health visits among Ontarian youth. Furthermore, there appears to be an upward trend of youth mental health utilization in the province of Ontario, especially noticeable in females who accessed primary health care services.

Incorporating Social Media into your Support Tool Box: Points to Consider from Genetics-Based Communities

Patients with newly-described or rare genetic findings are turning to social media to find and connect with others. Blogs, Facebook groups, and Twitter have all been reported as tools for patients to connect with one another. However, the preferences for social media use and privacy among patients, their families, and these communities have not been well characterized. To explore preferences about privacy and membership guidelines, an online survey was administered to two web-based patient registries, Simons Variation in Individuals Project ( www.simonsvipconnect.org ) and GenomeConnect ( www.genomeconnect.org ). Over a three-month period, invitations were sent to 2524 individuals and 103 responses (4%) were received and analyzed. Responses indicate that Facebook is the most popular resource accessed within this sample population (99%). Participants used social media to look for information about their diagnosis or test results (83%), read posts from rare disease groups or organizations (73%), participate in conversations about their diagnosis (67%), and connect with others to find support (58%). Focusing on privacy issues in social media, respondents indicate that membership and access impact the level of comfort in sharing personal or medical information. Nearly 60% of respondents felt uncomfortable sharing photos or medical information within a public Facebook group, whereas only 12% of respondents felt uncomfortable sharing in private group targeted to families alone. Using this preliminary data concerning social media use and privacy, we developed points for genetic counselors to incorporate when discussing available support resources for patients with a new, or rare, genetic diagnosis or genetic test result. Genetic counselors are trained to provide anticipatory guidance to families adapting to new genetic information, and are well-equipped to help patients consider their preferences about using social media as a source of information and support.

MicroRNAs in Cancer: challenges and opportunities in early detection, disease monitoring, and therapeutic agents

PURPOSE OF REVIEW

The goals of this review are to examine the usefulness of miRNAs as diagnostic and prognostic biomarkers for cancer and to evaluate the applicability of miRNAs as cancer therapeutics.

RECENT FINDINGS

Examination of miRNA milieu from body fluids offers a new alternative for quick, affordable and easy analysis of disease status in patients. Blood-based exosomal miRNAs have increased stability and are an excellent choice for clinical cancer diagnostics and prognostics. Currently, there are many miRNA signatures associated with cancer and progression but there is no consensus among multiple sera and tumor sample studies. Off-target and immunological effects remains an obstacle for use of miRNAs as novel chemotherapeutics in the clinic. Recent developments in nanotechnology and drug delivery systems which target the tumor microenvironment may provide an alternative therapeutic approach with decreased toxicity.

SUMMARY

This review critically evaluates the literature investigating the use of miRNAs as biomarkers and their future as potential therapeutics.