Testing the Impact of the #chatsafe Intervention on Young People's Ability to Communicate Safely About Suicide on Social Media: Protocol for a Randomized Controlled Trial

BACKGROUND

Suicide is the leading cause of death among Australians. One commonly cited explanation is the impact of social media, in particular, the ways in which young people use social media to communicate about their own experiences and their exposure to suicide-related content posted by others. Guidelines designed to assist mainstream media to safely report about suicide are widespread. Until recently, no guidelines existed that targeted social media or young people. In response, we developed the #chatsafe guidelines and a supporting social media campaign, which together make up the #chatsafe intervention. The intervention was tested in a pilot study with positive results. However, the study was limited by the lack of a control group.

OBJECTIVE

The aim of this study is to assess the impact of the #chatsafe social media intervention on young people's safety and confidence when communicating on the web about suicide.

METHODS

The study employs a pragmatic, parallel, superiority randomized controlled design. It will be conducted in accordance with the Consolidated Standards of Reporting Trials statement over 18 months. Participants will be 400 young people aged 16-25 years (200 per arm). Participants will be recruited via social media advertising and assessed at 3 time points: time 1-baseline; time 2-8-week postintervention commencement; and time 3-4-week postintervention. They will be asked to complete a weekly survey to monitor safety and evaluate each piece of social media content. The intervention comprises an 8-week social media campaign including social media posts shared on public Instagram profiles. The intervention group will receive the #chatsafe suicide prevention content and the control group will receive sexual health content. Both groups will receive 24 pieces of content delivered to their mobile phones via text message. The primary outcome is safety when communicating on the web about suicide, as measured via the purpose-designed #chatsafe online safety questionnaire. Additional outcomes include willingness to intervene against suicide, internet self-efficacy, safety, and acceptability.

RESULTS

The study was funded in November 2020, approved by the University of Melbourne Human Research Ethics Committee on October 7, 2022, and prospectively registered with the Australian New Zealand Clinical Trials registry. Trial recruitment began in November 2022 and study completion is anticipated by June 2024.

CONCLUSIONS

This will be the first randomized controlled trial internationally to test the impact of a social media intervention designed to equip young people to communicate safely on the web about suicide. Given the rising rates of youth suicide in Australia and the acceptability of social media among young people, incorporating social media-based interventions into the suicide prevention landscape is an obvious next step. This intervention, if effective, could also be extended internationally, thereby improving web-based safety for young people not just in Australia but globally.

TRIAL REGISTRATION

Australian New Zealand Clinical Trials Registry ACTRN12622001397707; https://anzctr.org.au/Trial/Registration/TrialReview.aspx?id=384318.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

DERR1-10.2196/44300.

Fairness in Mobile Phone–Based Mental Health Assessment Algorithms: Exploratory Study

Background

Approximately 1 in 5 American adults experience mental illness every year. Thus, mobile phone–based mental health prediction apps that use phone data and artificial intelligence techniques for mental health assessment have become increasingly important and are being rapidly developed. At the same time, multiple artificial intelligence–related technologies (eg, face recognition and search results) have recently been reported to be biased regarding age, gender, and race. This study moves this discussion to a new domain: phone-based mental health assessment algorithms. It is important to ensure that such algorithms do not contribute to gender disparities through biased predictions across gender groups.

Objective

This research aimed to analyze the susceptibility of multiple commonly used machine learning approaches for gender bias in mobile mental health assessment and explore the use of an algorithmic disparate impact remover (DIR) approach to reduce bias levels while maintaining high accuracy.

Methods

First, we performed preprocessing and model training using the data set (N=55) obtained from a previous study. Accuracy levels and differences in accuracy across genders were computed using 5 different machine learning models. We selected the random forest model, which yielded the highest accuracy, for a more detailed audit and computed multiple metrics that are commonly used for fairness in the machine learning literature. Finally, we applied the DIR approach to reduce bias in the mental health assessment algorithm.

Results

The highest observed accuracy for the mental health assessment was 78.57%. Although this accuracy level raises optimism, the audit based on gender revealed that the performance of the algorithm was statistically significantly different between the male and female groups (eg, difference in accuracy across genders was 15.85%; P<.001). Similar trends were obtained for other fairness metrics. This disparity in performance was found to reduce significantly after the application of the DIR approach by adapting the data used for modeling (eg, the difference in accuracy across genders was 1.66%, and the reduction is statistically significant with P<.001).

Conclusions

This study grounds the need for algorithmic auditing in phone-based mental health assessment algorithms and the use of gender as a protected attribute to study fairness in such settings. Such audits and remedial steps are the building blocks for the widespread adoption of fair and accurate mental health assessment algorithms in the future.

The Relationships of Deteriorating Depression and Anxiety With Longitudinal Behavioral Changes in Google and YouTube Use During COVID-19: Observational Study

BACKGROUND

Depression and anxiety disorders among the global population have worsened during the COVID-19 pandemic. Yet, current methods for screening these two issues rely on in-person interviews, which can be expensive, time-consuming, and blocked by social stigma and quarantines. Meanwhile, how individuals engage with online platforms such as Google Search and YouTube has undergone drastic shifts due to COVID-19 and subsequent lockdowns. Such ubiquitous daily behaviors on online platforms have the potential to capture and correlate with clinically alarming deteriorations in depression and anxiety profiles of users in a noninvasive manner.

OBJECTIVE

The goal of this study is to examine, among college students in the United States, the relationships of deteriorating depression and anxiety conditions with the changes in user behaviors when engaging with Google Search and YouTube during COVID-19.

METHODS

This study recruited a cohort of undergraduate students (N=49) from a US college campus during January 2020 (prior to the pandemic) and measured the anxiety and depression levels of each participant. The anxiety level was assessed via the General Anxiety Disorder-7 (GAD-7). The depression level was assessed via the Patient Health Questionnaire-9 (PHQ-9). This study followed up with the same cohort during May 2020 (during the pandemic), and the anxiety and depression levels were assessed again. The longitudinal Google Search and YouTube history data of all participants were anonymized and collected. From individual-level Google Search and YouTube histories, we developed 5 features that can quantify shifts in online behaviors during the pandemic. We then assessed the correlations of deteriorating depression and anxiety profiles with each of these features. We finally demonstrated the feasibility of using the proposed features to build predictive machine learning models.

RESULTS

Of the 49 participants, 49% (n=24) of them reported an increase in the PHQ-9 depression scores; 53% (n=26) of them reported an increase in the GAD-7 anxiety scores. The results showed that a number of online behavior features were significantly correlated with deteriorations in the PHQ-9 scores (r ranging between -0.37 and 0.75, all P values less than or equal to .03) and the GAD-7 scores (r ranging between -0.47 and 0.74, all P values less than or equal to .03). Simple machine learning models were shown to be useful in predicting the change in anxiety and depression scores (mean squared error ranging between 2.37 and 4.22, R2 ranging between 0.68 and 0.84) with the proposed features.

CONCLUSIONS

The results suggested that deteriorating depression and anxiety conditions have strong correlations with behavioral changes in Google Search and YouTube use during the COVID-19 pandemic. Though further studies are required, our results demonstrate the feasibility of using pervasive online data to establish noninvasive surveillance systems for mental health conditions that bypasses many disadvantages of existing screening methods.

Deploying nEmesis: Preventing Foodborne Illness by Data Mining Social Media

Foodborne illness afflicts 48 million people annually in the U.S. alone. Over 128,000 are hospitalized and 3,000 die from the infection. While preventable with proper food safety practices, the traditional restaurant inspection process has limited impact given the predictability and low frequency of inspections, and the dynamic nature of the kitchen environment. Despite this reality, the inspection process has remained largely unchanged for decades. CDC has even identified food safety as one of seven ”winnable battles”; however, progress to date has been limited. In this work, we demonstrate significant improvements in food safety by marrying AI and the standard inspection process. We apply machine learning to Twitter data, develop a system that automatically detects venues likely to pose a public health hazard, and demonstrate its efficacy in the Las Vegas metropolitan area in a double-blind experiment conducted over three months in collaboration with Nevada’s health department. By contrast, previous research in this domain has been limited to indirect correlative validation using only aggregate statistics. We show that adaptive inspection process is 64 percent more effective at identifying problematic venues than the current state of the art. If fully deployed, our approach could prevent over 9,000 cases of foodborne illness and 557 hospitalizations annually in Las Vegas alone. Additionally, adaptive inspections result in unexpected benefits, including the identification of venues lacking permits, contagious kitchen staff, and fewer customer complaints filed with the Las Vegas health department.

Blood Alcohol Concentrations in Suicide and Motor Vehicle Crash Decedents Ages 18 to 54

BACKGROUND

Using postmortem data, we examined the hypotheses that high (and very high) blood alcohol concentrations (BACs) are more common among motor vehicle crash decedents (MVCs) than among suicide decedents, whereas low alcohol levels are more common among suicides.

METHODS

We examined BAC in 224 suicide decedents and 166 MVCs ages 18 to 54 in the state of New Mexico in 2012. Comparisons between the groups were made based on differing BAC levels using 0.080 g/dl categories including low (0.001 to 0.079 g/dl), high (0.080 to 0.159 g/dl), and very high BAC (≥0.160 g/dl), and based on 0.100 g/dl categories including low (0.001 to 0.099 g/dl), high (0.100 to 0.199 g/dl), and very high BAC (≥0.200 g/dl), with these groups compared with a no-alcohol reference (0.000 g/dl) in separate analyses. Multivariate logistic regressions compared suicides with MVCs that adjusted for age, sex, and race/ethnicity.

RESULTS

Support for the hypothesis that suicides are more likely to have a low BAC level was supported in the analysis using the 0.100 g/dl categorizations. Neither analysis supported the hypothesis that MVCs are more likely to have high (or very high) BACs compared with suicides. Among both injury groups with positive BACs, low BACs were least common.

CONCLUSIONS

Low BAC levels may be more likely to be observed among suicides compared with MVCs, a possible reflection of the more varied role that alcohol plays in suicide compared with MVC. Nonetheless, high (and very high) BAC is the predominant scenario in both suicides and MVCs with positive BAC.

Suicide contagion: a systematic review of definitions and research utility

Objectives

Despite the common use of contagion to analogize the spread of suicide, there is a lack of rigorous assessment of the underlying concept or theory supporting the use of this term. The present study aims to examine the varied definitions and potential utility of the term contagion in suicide-related research.

Methods

100 initial records and 240 reference records in English were identified as relevant with our research objectives, through systematic literature screening. We then conducted narrative syntheses of various definitions and assessed their potential value for generating new research.

Results

20.3% of the 340 records used contagion as equivalent to clustering (contagion-as-cluster); 68.5% used it to refer to various, often related mechanisms underlying the clustering phenomenon (contagion-as-mechanism); and 11.2% without clear definition. Under the category of contagion-as-mechanism, four mechanisms have been proposed to explain how suicide clusters occurred: transmission (contagion-as-transmission), imitation (contagion-as-imitation), contextual influence (contagion-as-context), and affiliation (contagion-as-affiliation). Contagion-as-cluster both confounds and constrains inquiry into suicide clustering by blending proposed mechanism with the phenomenon to be studied. Contagion-as-transmission is, in essence, a double or internally redundant metaphor. Contagion-as-affiliation and contagion-as-context involve mechanisms that are common mechanisms that often occur independently of apparent contagion, or may serve as a facilitating background. When used indiscriminately, these terms may create research blind spots. Contagion-as-imitation combines perspectives from psychology, sociology, and public health research and provides the greatest heuristic utility for examining whether and how suicide and suicidal behaviors may spread among persons at both individual and population levels.

Conclusion

Clarifying the concept of “suicide contagion” is an essential step for more thoroughly investigating its mechanisms. Developing a clearer understanding of the apparent spread of suicide-promoting influences can, in turn, offer insights necessary to build the scientific foundation for prevention and intervention strategies that can be applied at both individual and community levels.

Health care utilization and health indicators among a national sample of U.S. veterans in same-sex partnerships

OBJECTIVES

To examine health indicators of same-sex partnered veterans as compared with their opposite-sex partnered veteran and nonveteran peers.

METHODS

Same-sex partner status was derived by self-reported same-sex partnerships in data from the 2004 Behavioral Risk Factor Surveillance System. Outcome variables included health risk disparities associated with sexual minority status (e.g., frequent mental distress) and veteran status (e.g., firearm ownership). Stratified multiple logistic regression models were used to examine the association of same-sex partnered veteran status with health indicators.

RESULTS

Same-sex partnered veterans had higher odds of being overweight and keeping firearms in the house compared with same-sex partnered nonveterans. Same-sex partnered veterans were less likely than opposite-sex partnered veterans to be overweight, and they were more than twice as likely to be current smokers when compared with opposite-sex partnered nonveterans.

CONCLUSIONS

Findings suggest both that some health disparities patterns identified by same-sex partnership status among the general population also exist among veteran populations, and that some unique distinctions may exist, particularly related to BMI and firearm ownership. Collection of information about sexual minority status within Department of Veterans Affairs data sources is needed to more accurately assess the health of this minority population.

Use of personal digital assistants to enhance educational evaluation in a primary care clerkship

Experiences of students using optically scanned cards were compared with those of students using personal digital assistants (PDAs) to log patient encounters in a primary care clerkship. From April to September 2001, students were offered the option of using a PDA in lieu of scanned cards to track clinical encounters. Data obtained from PDA users were compared with those previously obtained from scanned card users. Verbal and written feedback was obtained from all students. Of the 71 students invited to participate, 21 (30%) owned a PDA, and of these, 20 agreed to participate. Eighteen students completed the pilot. One student was unable to participate owing to software installation problems; another student lost data because of improper back-up. Students using the PDAs recorded more encounters per rotation and had fewer missing data when compared with students who used scanned card. Additionally, feedback from students suggested that PDAs offered other important educational benefits.

Detection of E. coli colonies expressing the v-sis oncogene product with monoclonal antibodies made against synthetic peptides

A method for immunodetection of individual epitopes on eukaryotic proteins synthesized in E. coli colonies is described. The system is developed using monoclonal antibodies produced against the Ha-ras protein produced in E. coli JM103. Monoclonal antibodies made against a synthetic peptide from the v-sis oncogene sequence are then used to identify bacterial colonies in which the v-sis protein is being produced. Production of the v-sis protein by these E. coli colonies was confirmed by immunoblot analysis. The assay utilizes peroxidase conjugated anti-mouse immunoglobulin and 4-chloro-1-naphthol to detect the positive colonies and can detect on the order of 200 pg antigen per E. coli colony.