Understanding and detecting behaviours prior to a suicide attempt: A mixed-methods study.

OBJECTIVE: Prior research suggests there are observable behaviours preceding suicide attempts in public places. However, there are currently no ways to continually monitor such sites, limiting the potential to intervene. In this mixed-methods study, we examined the acceptability and feasibility of using an automated computer system to identify crisis behaviours. METHODS: First, we conducted a large-scale acceptability survey to assess public perceptions on research using closed-circuit television and artificial intelligence for suicide prevention. Second, we identified crisis behaviours at a frequently used cliff location by manual structured analysis of closed-circuit television footage. Third, we configured a computer vision algorithm to identify crisis behaviours and evaluated its sensitivity and specificity using test footage. RESULTS: Overall, attitudes were positive towards research using closed-circuit television and artificial intelligence for suicide prevention, including among those with lived experience. The second study revealed that there are identifiable behaviours, including repetitive pacing and an extended stay. Finally, the automated behaviour recognition algorithm was able to correctly identify 80% of acted crisis clips and correctly reject 90% of acted non-crisis clips. CONCLUSION: The results suggest that using computer vision to detect behaviours preceding suicide is feasible and well accepted by the community and may be a feasible method of initiating human contact during a crisis.

Towards Building a Visual Behaviour Analysis Pipeline for Suicide Detection and Prevention.

Understanding human behaviours through video analysis has seen significant research progress in recent years with the advancement of deep learning. This topic is of great importance to the next generation of intelligent visual surveillance systems which are capable of real-time detection and analysis of human behaviours. One important application is to automatically monitor and detect individuals who are in crisis at suicide hotspots to facilitate early intervention and prevention. However, there is still a significant gap between research in human action recognition and visual video processing in general, and their application to monitor hotspots for suicide prevention. While complex backgrounds, non-rigid movements of pedestrians and limitations of surveillance cameras and multi-task requirements for a surveillance system all pose challenges to the development of such systems, a further challenge is the detection of crisis behaviours before a suicide attempt is made, and there is a paucity of datasets in this area due to privacy and confidentiality issues. Most relevant research only applies to detecting suicides such as hangings or jumps from bridges, providing no potential for early prevention. In this research, these problems are addressed by proposing a new modular design for an intelligent visual processing pipeline that is capable of pedestrian detection, tracking, pose estimation and recognition of both normal actions and high risk behavioural cues that are important indicators of a suicide attempt. Specifically, based on the key finding that human body gestures can be used for the detection of social signals that potentially precede a suicide attempt, a new 2D skeleton-based action recognition algorithm is proposed. By using a two-branch network that takes advantage of three types of skeleton-based features extracted from a sequence of frames and a stacked LSTM structure, the model predicts the action label at each time step. It achieved good performance on both the public dataset JHMDB and a smaller private CCTV footage collection on action recognition. Moreover, a logical layer, which uses knowledge from a human coding study to recognise pre-suicide behaviour indicators, has been built on top of the action recognition module to compensate for the small dataset size. It enables complex behaviour patterns to be recognised even from smaller datasets. The whole pipeline has been tested in a real-world application of suicide prevention using simulated footage from a surveillance system installed at a suicide hotspot, and preliminary results confirm its effectiveness at capturing crisis behaviour indicators for early detection and prevention of suicide.

Ethics and Privacy in Social Media Research for Mental Health.

PURPOSE OF REVIEW: This review provides an overview of recent research which uses social media data in the context of mental health. It also provides an overview of challenges in relation to consent, privacy, and usage of such data. RECENT FINDINGS: A broad range of research has been conducted in recent years, using text-based and visual data from social media platforms, for purposes such as risk detection at the individual level, providing crisis outreach, and developing a better understanding of the lived experience of mental ill-health. Challenges remain in relation to obtaining truly informed consent for research using social media data-however platforms allowing data donation may address these concerns. There is an imperative need to ensure that privacy is preserved at all stages of the research process, from data collection, to analysis, and the responsible use of raw data in publications.

Attitudes among the Australian public toward AI and CCTV in suicide prevention research: A mixed methods study.

Research is underway exploring the use of closed-circuit television (CCTV) cameras and artificial intelligence (AI) for suicide prevention research in public locations where suicides occur. Given the sensitive nature and potential implications of this research, this study explored ethical concerns the public may have about research of this nature. Developed based on the principle of respect, a survey was administered to a representative sample of 1,096 Australians to understand perspectives on the research. The sample was aged 18 and older, 53% female, and 9% ethnic minority. Following an explanatory mixed methods approach, interviews and a focus group were conducted with people with a lived experience of suicide and first responders to contextualize the findings. There were broad levels of acceptance among the Australian public. Younger respondents, females, and those declining to state their ethnicity had lower levels of acceptance of CCTV research using AI for suicide prevention. Those with lived experience of suicide had higher acceptance. Qualitative data indicated concern regarding racial bias in AI and police response to suicidal crises and the need for lived experience involvement in the development and implementation of any resulting interventions. Broad public acceptance of the research aligns with the principle of respect for persons. Beneficence emerged in the context of findings emphasizing the importance of meaningfully including people with lived experience in the development and implementation of interventions resulting from this research, while justice emerged in themes expressing concerns about racial bias in AI and police response to mental health crises. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

The Effect of Explicit Suicide Language in Engagement With a Suicide Prevention Search Page Help-Seeking Prompt: Nonrandomized Trial.

BACKGROUND: Given that signage, messaging, and advertisements (ads) are the gateway to many interventions in suicide prevention, it is important that we understand what type of messaging works best for whom. OBJECTIVE: We investigated whether explicitly mentioning suicide increases engagement using internet ads by investigating engagement with campaigns with different categories of keywords searched, which may reflect different cognitive states. METHODS: We ran a 2-arm study Australia-wide, with or without ads featuring explicit suicide wording. We analyzed whether there were differences in engagement for campaigns with explicit and nonexplicit ads for low-risk (distressed but not explicitly suicidal), high-risk (explicitly suicidal), and help-seeking for suicide keywords. RESULTS: Our analyses revealed that having explicit wording has opposite effects, depending on the search terms used: explicit wording reduced the engagement rate for individuals searching for low-risk keywords but increased engagement for those using high-risk keywords. CONCLUSIONS: The findings suggest that individuals who are aware of their suicidality respond better to campaigns that explicitly use the word "suicide." We found that individuals who search for low-risk keywords also respond to explicit ads, suggesting that some individuals who are experiencing suicidality search for low-risk keywords.

Indonesia's first suicide statistics profile: an analysis of suicide and attempt rates, underreporting, geographic distribution, gender, method, and rurality.

Background: Timely and accurate data are critical for effective suicide prevention. Indonesia-the fourth most populous country in the world-has limited data availability and thus, limited data-driven interventions. Through a national government partnership, we obtained critical non-public data for attempts and suicides that could be analysed for the first time in Indonesia's history. Methods: We obtained and analysed data from five sources from 2016 to 2021: police data, death registry data, a provincial survey, a sample registry system, and the WHO's Global Health Observatory (WHO GHO) data. Using these data, we estimated underreporting, identified provinces with the highest suicide and suicide attempt rates, assessed gender ratios, identified methods used, and compared urban and rural suicides. Findings: The analysis yielded an underreporting rate of 859.10% for suicides, while verbal autopsies and increased quality control only increased coverage from 12.80% to 51.40%. Provinces with the highest rates of suicide were Bali, Riau Islands, Special Region of Yogyakarta, Central Java, and Central Kalimantan. Gender analysis revealed a ratio of 1: 2.11 for female to male suicides. Suicide methods analysis revealed that hanging and self-poisoning were the most used method, and rural suicides occurred at a rate 4.47 times higher than urban suicides. Interpretation: The analysis revealed the highest underreporting rate in the literature from a national sample and vast heterogeneity among provinces with high suicide rates-including provinces with strong mystic beliefs, suggesting the need for a culturally sensitive sub-national tailored approach. Through our study, we provide critical information which will allow for data-driven suicide prevention. Funding: The data collection for this was part of a project funded by the Australian Department of Foreign Affairs and Trade, Australian-Indonesian Institute (AII2020322).

Images of Australian alcoholic and non-alcoholic beverages: A validation dataset.

For multi-session alcohol cognitive bias modification, a large image dataset depicting both alcohol and non-alcoholic beverages is required. We photographed a wide range of beverages and then validated them in a group of Australian community participants: 47 women and 39 men, aged from 18 to 73, who drank alcohol at least occasionally in the last year, with Alcohol Use Disorder Identification Test (AUDIT) scores ranging from 1 to 33. Participants were asked to categorize images as alcoholic vs non-alcoholic, rate the familiarity of each beverage and rate their craving for each beverage. The dataset includes all images and ratings for each image, stratified by gender and high/low AUDIT scores. Mean ratings per participant per beverage category are also provided.

Gone for good: Lack of priming suggests early perceptual interference in emotion-induced blindness with negative stimuli.

Emotionally negative stimuli are perceptually prioritized to such a degree that they can cause people to miss seeing subsequent targets that appear in front of their eyes. It is unclear whether this effect (known as emotion-induced blindness) reflects postperceptual interference, in which case unseen targets might still impact later responses, as in the seemingly similar "attentional blink". An alternative is that emotional distractors prevent target encoding, and so leave no residual trace of target information. In this study, we used a priming task to assess these alternative possibilities. Each emotion-induced blindness trial was immediately followed by a speeded arrow judgment task, in which the arrow's orientation could be congruent or incongruent with the orientation of an emotion-induced blindness target. Analyses revealed strong evidence that seen targets primed the arrow judgment, but there was moderate to strong evidence that unseen targets elicited no priming whatsoever. These results lend support to claims that emotion-induced blindness reflects failure to perceptually encode target information, and may reflect a different mechanism from the phenomenally similar attentional blink. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Evaluating a Human Detection Model in a Behaviour Analysis Pipeline for Suicide Prevention.

Suicides in public places, such as railways, can have a significant impact on bystanders, railway staff, first responders and the surrounding communities. Behaviours prior to a suicide attempt have been identified, that could potentially be detected automatically. As a first step, the algorithm is required to accurately identify individuals exhibiting these behaviours in different settings. Our study analyses a human detection model focussing on pedestrian detection at railway stations as one component of a broader project to detect pre-suicidal behaviours. Closed-circuit television footage from two stations collected for the same 24-hour period were manually analysed to obtain parameters (true positives, false positives, and false negatives) which were then used to compute performance measures (sensitivity, precision, and F1 score). The model performed differently in both stations with a sensitivity of 0.73 and F1 score of 0.84 in Station A and a sensitivity of 0.48 and F1 score of 0.65 in Station B. Root causes of false negatives identified include differing body postures and occlusion. Although the model was adequate, its performance is dependent on the view captured by the cameras in stations. Collectively, these findings can be used to improve the model's performance.Clinical Relevance-Detecting behaviours prior to a suicide attempt offers a critical period for intervention by bystanders or first responders, potentially interrupting the attempt. This offers the potential to directly reduce suicide attempts, as well as reduce third-party exposure to these traumatic events.

Suicide Prevention Using Google Ads: Randomized Controlled Trial Measuring Engagement.

BACKGROUND: Studies have shown that individuals may search for suicide-related terms on the internet prior to an attempt. OBJECTIVE: Thus, across 2 studies, we investigated engagement with an advertisement campaign designed to reach individuals contemplating suicide. METHODS: First, we designed the campaign to focus on crisis, running a campaign for 16 days in which crisis-related keywords would trigger an ad and landing page to help individuals find the national suicide hotline number. Second, we expanded the campaign to also help individuals contemplating suicide, running the campaign for 19 days with a wider range of keywords through a co-designed website with a wider range of offerings (eg, lived experience stories). RESULTS: In the first study, the ad was shown 16,505 times and was clicked 664 times (4.02% click rate). There were 101 calls to the hotline. In the second study, the ad was shown 120,881 times and clicked 6227 times (5.15% click rate); of these 6227 clicks, there were 1419 (22.79%) engagements with the site, a substantially higher rate than the industry average of 3%. The number of clicks on the ad was high despite a suicide hotline banner likely being present. CONCLUSIONS: Search advertisements are a quick, far-reaching, and cost-efficient way of reaching those contemplating suicide and are needed despite suicide hotline banners being present. TRIAL REGISTRATION: Australian New Zealand Clinical Trials Registry (ANZCTR) ACTRN12623000084684; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=385209.

On the relative sensitivity of spatial and nonspatial measures of attentional bias: Emotion-induced blindness, the dot probe, and gradations in ratings of negative pictures.

Attentional biases toward negative information are implicated in various emotional disorders. The literature probing this relationship relies on assumptions that the tasks used to measure attentional biases are sensitive to the negative emotional qualities of stimuli; but are such assumptions justified? We assessed the degree to which two widely used tasks-the dot probe and emotion-induced blindness-displayed sensitivity to gradations in valence and arousal ratings for negative emotional pictures. For emotion-induced blindness (the failure to see a target that follows an emotional distractor in a rapidly presented sequence of items), there was strong evidence of sensitivity to gradations in both valence and arousal. In contrast, there was moderate to strong evidence that the dot probe (a spatial attention task where latency to respond to a target depends on whether it appears at or away from the location of an emotional stimulus) was insensitive to gradations in valence and arousal, although there was some evidence of its sensitivity to emotional versus neutral stimuli overall. That said, in the dot probe, response latency regardless of spatial relationship between the target and the emotional image appeared sensitive to gradations in stimulus emotionality. Together, these findings suggest that such sensitivity may be characteristic of nonspatial, rather than spatial, aspects of attention. Implications for attentional bias studies are discussed. Notably, the finding that emotion-induced blindness was sensitive to gradations in ratings of emotional pictures supports claims that the effect arises due to stimulus emotionality rather than simply differences in visual features of pictures (e.g., color, brightness, complexity). (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Blinded by and Stuck in Negative Emotions: Is Psychological Inflexibility Across Different Domains Related?

Psychological inflexibility is theorized to underlie difficulties adjusting mental processes in response to changing circumstances. People show inflexibility across a range of domains, including attention, cognition, and affect. But it remains unclear whether common mechanisms underlie inflexibility in different domains. We investigated this possibility in a pre-registered replication and extension examining associations among attentional, cognitive, and affective inflexibility measures. Participants (N = 196) completed lab tasks assessing (a) emotion-induced blindness, the tendency for task-irrelevant emotional stimuli to impair attention allocation to non-emotional stimuli; (b) emotional inertia, the tendency for feelings to persist across time and contexts; and global self-report measures of (c) repetitive negative thinking, the tendency to repeatedly engage in negative self-focused thoughts (i.e., rumination, worry). Based on prior research linking repetitive negative thinking with negative affect inertia, on one hand, and emotion-induced blindness, on the other, we predicted positive correlations among all three measures of inflexibility. However, none of the three measures were related and Bayes factors indicated strong evidence for independence. Supplementary analyses ruled out alternative explanations for our findings, e.g., analytic decisions. Although our findings question the overlap between attentional, cognitive, and affective inflexibility measures, this study has methodological limitations. For instance, our measures varied across more than their inflexibility domain and our sample, relative to previous studies, included a high proportion of Asian participants who may show different patterns of ruminative thinking to non-Asian participants. Future research should address these limitations to confirm that common mechanisms do not underlie attentional, cognitive, and affective inflexibility. Supplementary Information: The online version contains supplementary material available at 10.1007/s42761-022-00145-2.

Cost-effectiveness of Installing Barriers at Bridge and Cliff Sites for Suicide Prevention in Australia.

Importance: Installation of barriers has been shown to reduce suicides. To our knowledge, no studies have evaluated the cost-effectiveness of installing barriers at multiple bridge and cliff sites where suicides are known to occur. Objective: To examine the cost-effectiveness of installing barriers at bridge and cliff sites throughout Australia. Design, Setting, and Participants: This economic evaluation used an economic model to examine the costs, costs saved, and reductions in suicides if barriers were installed across identified bridge and cliff sites over 5 and 10 years. Specific and accessible bridge and cliff sites across Australia that reported 2 or more suicides over a 5-year period were identified for analysis. A partial societal perspective (including intervention costs and monetary value associated with preventing suicide deaths) was adopted in the development of the model. Interventions: Barriers installed at bridge and cliff sites. Main Outcomes and Measures: Primary outcome was return on investment (ROI) comparing cost savings with intervention costs. Secondary outcomes included incremental cost-effectiveness ratio (ICER), comprising the difference in costs between installation of barriers and no installation of barriers divided by the difference in reduction of suicide cases. Uncertainty and sensitivity analyses were undertaken to examine the association of changes in suicide rates with barrier installation, adjustments to the value of statistical life, and changes in maintenance costs of barriers. Results: A total of 7 bridges and 19 cliff sites were included in the model. If barriers were installed at bridge sites, an estimated US $145 million (95% uncertainty interval [UI], $90 to $160 million) could be saved in prevented suicides over 5 years, and US $270 million (95% UI, $176 to $298 million) over 10 years. The estimated ROI ratio for building barriers over 10 years at bridges was 2.4 (95% UI, 1.5 to 2.7); the results for cliff sites were not significant (ROI, 2.0; 95% UI, -1.1 to 3.8). The ICER indicated monetary savings due to averted suicides over the intervention cost for bridges, although evidence for similar savings was not significant for cliffs. Results were robust in all sensitivity analyses except when the value of statistical life-year over 5 or 10 years only was used. Conclusions and Relevance: In an economic analysis, barriers were a cost-effective suicide prevention intervention at bridge sites. Further research is required for cliff sites.

The Use of Closed-Circuit Television and Video in Suicide Prevention: Narrative Review and Future Directions.

BACKGROUND: Suicide is a recognized public health issue, with approximately 800,000 people dying by suicide each year. Among the different technologies used in suicide research, closed-circuit television (CCTV) and video have been used for a wide array of applications, including assessing crisis behaviors at metro stations, and using computer vision to identify a suicide attempt in progress. However, there has been no review of suicide research and interventions using CCTV and video. OBJECTIVE: The objective of this study was to review the literature to understand how CCTV and video data have been used in understanding and preventing suicide. Furthermore, to more fully capture progress in the field, we report on an ongoing study to respond to an identified gap in the narrative review, by using a computer vision-based system to identify behaviors prior to a suicide attempt. METHODS: We conducted a search using the keywords "suicide," "cctv," and "video" on PubMed, Inspec, and Web of Science. We included any studies which used CCTV or video footage to understand or prevent suicide. If a study fell into our area of interest, we included it regardless of the quality as our goal was to understand the scope of how CCTV and video had been used rather than quantify any specific effect size, but we noted the shortcomings in their design and analyses when discussing the studies. RESULTS: The review found that CCTV and video have primarily been used in 3 ways: (1) to identify risk factors for suicide (eg, inferring depression from facial expressions), (2) understanding suicide after an attempt (eg, forensic applications), and (3) as part of an intervention (eg, using computer vision and automated systems to identify if a suicide attempt is in progress). Furthermore, work in progress demonstrates how we can identify behaviors prior to an attempt at a hotspot, an important gap identified by papers in the literature. CONCLUSIONS: Thus far, CCTV and video have been used in a wide array of applications, most notably in designing automated detection systems, with the field heading toward an automated detection system for early intervention. Despite many challenges, we show promising progress in developing an automated detection system for preattempt behaviors, which may allow for early intervention.

Learned value and predictiveness affect gaze but not figure assignment.

Many factors affect figure-ground segregation, but the contributions of attention and reward history to this process is uncertain. We conducted two experiments to investigate whether reward learning influences figure assignment and whether this relationship was mediated by attention. Participants learned to associate certain shapes with a reward contingency: During a learning phase, they chose between two shapes on each trial, with subsets of shapes associated with high-probability win, low-probability win, high-probability loss, and low-probability loss. In a test phase, participants were given a figure-ground task, in which they indicated which of two regions that shared a contour they perceived as the figure (high-probability win and low-probability win shapes were pitted against each other, as were high-probability loss and low-probability loss shapes). The results revealed that participants had learned the reward contingencies and that, following learning, attention was reliably drawn to the optimal stimulus. Despite this, neither reward history nor the resulting attentional allocation influenced figure-ground organization.

Is there evidence for cross-domain congruency sequence effect? A replication of Kan et al. (2013).

Exploring the mechanisms of cognitive control is central to understanding how we control our behaviour. These mechanisms can be studied in conflict paradigms, which require the inhibition of irrelevant responses to perform the task. It has been suggested that in these tasks, the detection of conflict enhances cognitive control resulting in improved conflict resolution of subsequent trials. If this is the case, then this so-called congruency sequence effect can be expected to occur in cross-domain tasks. Previous research on the domain-generality of the effect presented inconsistent results. In this study, we provide a multi-site replication of three previous experiments of Kan et al. (Kan IP, Teubner-Rhodes S, Drummey AB, Nutile L, Krupa L, Novick JM 2013 Cognition 129, 637-651) which test congruency sequence effect between very different domains: from a syntactic to a non-syntactic domain (Experiment 1), and from a perceptual to a verbal domain (Experiments 2 and 3). Despite all our efforts, we found only partial support for the claims of the original study. With a single exception, we could not replicate the original findings; the data remained inconclusive or went against the theoretical hypothesis. We discuss the compatibility of the results with alternative theoretical frameworks.