Abstract Background Schizophrenia is a serious chronic mental disorder. Patients often show positive symptoms such as hallucinations and delusions, as well as negative symptoms such as emotional apathy and social withdrawal. Art therapy, with its advantages such as non-invasiveness and humanistic care, has gradually become an important supplementary means of mental health care. However, most existing research focuses on the application of broad art therapy and lacks customized intervention exploration based on the pathological characteristics of schizophrenia patients. Based on this, the research aims to verify the actual effect of art therapy in improving patients’ mental symptoms through quantitative data analysis. Methods A randomized controlled trial was conducted to select 60 patients with stable schizophrenia from a certain mental health center as the research subjects. They were randomly and equally divided into the intervention group and the control group. There were no statistically significant differences between the two groups in terms of gender, age, disease duration, and core scale scores before intervention (p>.05), and they were comparable. Among them, the intervention group received 12 weeks of art design intervention on the basis of routine care, color perception and emotional expression training from 1 to 4 weeks, three-dimensional handicraft creation and cognitive reconstruction from 5 to 8 weeks, and therapeutic space interaction and social practice from 9 to 12 weeks, while the control group only received routine rehabilitation care. The research was conducted based on the Positive and Negative Syndrome Scale (PANSS) and the Social Disability Screening Schedule. The main indicators were SDSS and Serum Cortisol Detection (SCD). Data were collected from the two groups of patients before the intervention (T0) and after the intervention (T1), respectively. Paired t-tests and independent sample t-tests were performed using SPSS 26.0 software. Results The specific experimental results are shown in Table 1. It can be seen from Table 1 that all the indicators of the intervention group are significantly better than those of the control group. Among them, the total PANSS score of the intervention group decreased from 89.65 ± 10.31 in the T0 stage to 62.34 ± 8.10 after art therapy, with a decrease of 30.46%, and it was statistically significant (p<.001). The decrease range of SDSS score (54.02%) was significantly higher than that of the control group (26.12%), while SCD decreased to 385.68 ± 45.20 after the intervention, which was significantly lower than 497.85 ± 56.79 of the control group. Moreover, both SDSS and SCD in the intervention group were statistically significant (p<.001). Discussion Research has confirmed that immersive art design intervention has a significant therapeutic effect on patients with stable schizophrenia. Its therapeutic mechanism involves physiological, psychological and social dimensions, providing a scalable solution for the rehabilitation care of schizophrenia patients.
The widespread adoption of generative artificial intelligence conversational agents (AICAs) among university students constitutes a novel cognitive social environment whose impact on the maturing brain remains elusive. Combining surveys with high resolution structural MRI, we examined patterns of general, functional, and socio emotional AICA use, academic performance, mental health, and brain structural signatures in a comparatively large sample of 222 young individuals. Across computational anatomy, meta analytic network level, and behavioral decoding analyses, we observed use specific associations. Higher general and functional AICA use frequencies were linked to better academic outcomes (GPA), larger dorsolateral prefrontal and calcarine gray matter volume, and enhanced hippocampal network clustering and local efficiency. In contrast, more frequent socio emotional AICA use was associated with poorer mental health (depression, social anxiety) and lower volume of superior temporal and amygdalar regions central to social and affective processing. These findings indicate that the same class of AI tools exerts distinct effects depending on usage patterns and motivations, engaging prefrontal hippocampal systems that support cognition versus socio emotional systems that may track distress linked usage. These heterogeneities are crucial for designing environments that harness the educational benefits of AI while mitigating mental health risks.
Text-based conversational agents (CAs) are increasingly used in mental health, yet evaluation practices remain fragmented. We conducted a PRISMA-guided systematic review (May-June 2024) across ACM Digital Library, Scopus, and PsycINFO. From 613 records, 132 studies were included, with dual-coder extraction achieving substantial agreement (Cohen's kappa = 0.77-0.92). We synthesized evaluation approaches across three dimensions: metrics, methods, and usage contexts. Metrics were classified into CA-centric attributes (e.g., reliability, safety, empathy) and user-centric outcomes (experience, knowledge, psychological state, health behavior). Methods included automated analyses, standardized psychometric scales, and qualitative inquiry. Temporal designs ranged from momentary to follow-up assessments. Findings show reliance on Western-developed scales, limited cultural adaptation, predominance of small and short-term samples, and weak links between automated performance metrics and user well-being. We argue for methodological triangulation, temporal rigor, and equity in measurement. This review offers a structured foundation for reliable, safe, and user-centered evaluation of mental health CAs.
Veith Weilnhammer, Kevin YC Hou, Lennart Luettgau
et al.
Millions of users turn to consumer AI chatbots to discuss mental health and behavioral concerns. While this presents unprecedented opportunities to deliver population-level support, it also highlights an urgent need for rigorous and scalable safety evaluations. Here we introduce SIM-VAIL, an AI chatbot auditing framework that captures how harmful chatbot responses manifest across a range of mental health contexts. SIM-VAIL pairs a simulated user, harboring a distinct psychiatric vulnerability and conversational intent, with a frontier AI chatbot. It scores conversation turns on 13 clinically relevant risk dimensions, enabling context-dependent, temporally resolved safety assessment. Across 810 conversations, encompassing over 90,000 turn-level ratings and 30 psychiatric user profiles, we found evidence of concerning chatbot behavior across virtually all user phenotypes and most of the 9 consumer AI chatbots audited, albeit reduced in newer models. Rather than arising abruptly, concerning behavior accumulated over multiple turns. Risk profiles were phenotype-dependent and exhibited trade-offs, indicating that chatbot behaviors that appear supportive in general settings can become maladaptive when they align with mechanisms that sustain a user's vulnerability. These findings identify a systematic failure mode in human-AI interactions, which we term Vulnerability-Amplifying Interaction Loops (VAILs), and underscore the need for multidimensional approaches to risk quantification. SIM-VAIL provides a scalable framework for quantifying how mental health risk is distributed across user phenotypes, conversational trajectories, and clinically grounded behavioral dimensions, offering a new foundation for targeted safety improvements.
Andrea Cuadra, Samar Sabie, Yan Shvartzshnaider
et al.
We investigate the ethical and privacy implications of voice-first ambient interfaces (VFAIs) for aging in place through an in-depth engagement with five older adults. Our participants were in the process of becoming experienced VFAI users, and had used a VFAI-based design probe for health data reporting. We create and iteratively refine an interview protocol using Privacy Cards. We customize Privacy Cards by drawing on participants' previous interviews and device usage logs. Using Privacy Cards, we conduct interviews to surface their mental models, and explore their privacy concerns. We find insufficient mental models for proper consent. For example, participants did not know who could access their data, and experienced difficulty distinguishing built-in functionality from third-party apps. Participants initially expressed little worry about VFAI-related ethical concerns, but interviews with Privacy Cards revealed nuanced issues, resulting in various implications for future research and design.
Large language models (LLMs) are increasingly deployed as tool-using agents, shifting safety concerns from harmful text generation to harmful task completion. Deployed systems often condition on user profiles or persistent memory, yet agent safety evaluations typically ignore personalization signals. To address this gap, we investigated how mental health disclosure, a sensitive and realistic user-context cue, affects harmful behavior in agentic settings. Building on the AgentHarm benchmark, we evaluated frontier and open-source LLMs on multi-step malicious tasks (and their benign counterparts) under controlled prompt conditions that vary user-context personalization (no bio, bio-only, bio+mental health disclosure) and include a lightweight jailbreak injection. Our results reveal that harmful task completion is non-trivial across models: frontier lab models (e.g., GPT 5.2, Claude Sonnet 4.5, Gemini 3-Pro) still complete a measurable fraction of harmful tasks, while an open model (DeepSeek 3.2) exhibits substantially higher harmful completion. Adding a bio-only context generally reduces harm scores and increases refusals. Adding an explicit mental health disclosure often shifts outcomes further in the same direction, though effects are modest and not uniformly reliable after multiple-testing correction. Importantly, the refusal increase also appears on benign tasks, indicating a safety--utility trade-off via over-refusal. Finally, jailbreak prompting sharply elevates harm relative to benign conditions and can weaken or override the protective shift induced by personalization. Taken together, our results indicate that personalization can act as a weak protective factor in agentic misuse settings, but it is fragile under minimal adversarial pressure, highlighting the need for personalization-aware evaluations and safeguards that remain robust across user-context conditions.
We study the psychological and social impacts of exiting a national anti-poverty campaign, leveraging China's phase-out of its national poverty assistance as a natural experiment. Using a regression discontinuity design, we find that exiting the national campaign improves mental wellbeing. These improvements are accompanied by stronger social and family ties -- such as greater perceived support and communication, while income and material conditions remain largely unchanged. Our findings offer insights into the design of policy exits and underscore the importance of incorporating measures that sustain community- and family-based support systems when implementing or ending assistance programs.
In the context of explainable artificial intelligence (XAI), limited research has identified role-specific explanation needs. This study investigates the explanation needs of data scientists, who are responsible for training, testing, deploying, and maintaining machine learning (ML) models in AI systems. The research aims to determine specific explanation content of data scientists. A task analysis identified user goals and proactive user tasks. Using explanation questions, task-specific explanation needs and content were identified. From these individual explanations, we developed a mental model for explanations, which was validated and revised through a qualitative study (n=12). In a second quantitative study (n=12), we examined which explanation intents (reason, comparison, accuracy, prediction, trust) require which type of explanation content from the mental model. The findings are: F1: Explanation content for data scientists comes from the application domain, system domain, and AI domain. F2: Explanation content can be complex and should be organized sequentially and/or in hierarchies (novelty claim). F3: Explanation content includes context, inputs, evidence, attributes, ranked list, interim results, efficacy principle, and input/output relationships (novelty claim). F4: Explanation content should be organized as a causal story. F5: Standardized explanation questions ensure complete coverage of explanation needs (novelty claim). F6: Refining mental models for explanations increases significantly its quality (novelty claim).
Should a large language model (LLM) be used as a therapist? In this paper, we investigate the use of LLMs to *replace* mental health providers, a use case promoted in the tech startup and research space. We conduct a mapping review of therapy guides used by major medical institutions to identify crucial aspects of therapeutic relationships, such as the importance of a therapeutic alliance between therapist and client. We then assess the ability of LLMs to reproduce and adhere to these aspects of therapeutic relationships by conducting several experiments investigating the responses of current LLMs, such as `gpt-4o`. Contrary to best practices in the medical community, LLMs 1) express stigma toward those with mental health conditions and 2) respond inappropriately to certain common (and critical) conditions in naturalistic therapy settings -- e.g., LLMs encourage clients' delusional thinking, likely due to their sycophancy. This occurs even with larger and newer LLMs, indicating that current safety practices may not address these gaps. Furthermore, we note foundational and practical barriers to the adoption of LLMs as therapists, such as that a therapeutic alliance requires human characteristics (e.g., identity and stakes). For these reasons, we conclude that LLMs should not replace therapists, and we discuss alternative roles for LLMs in clinical therapy.
Giuseppe Desolda, Andrea Esposito, Francesco Greco
et al.
Integrated Development Environments increasingly implement AI-powered code completion tools (CCTs), which promise to enhance developer efficiency, accuracy, and productivity. However, interaction challenges with CCTs persist, mainly due to mismatches between developers' mental models and the unpredictable behavior of AI-generated suggestions, which is an aspect underexplored in the literature. We conducted an elicitation study with 56 developers using co-design workshops to elicit their mental models when interacting with CCTs. Different important findings that might drive the interaction design with CCTs emerged. For example, developers expressed diverse preferences on when and how code suggestions should be triggered (proactive, manual, hybrid), where and how they are displayed (inline, sidebar, popup, chatbot), as well as the level of detail. It also emerged that developers need to be supported by customization of activation timing, display modality, suggestion granularity, and explanation content, to better fit the CCT to their preferences. To demonstrate the feasibility of these and the other guidelines that emerged during the study, we developed ATHENA, a proof-of-concept CCT that dynamically adapts to developers' coding preferences and environments, ensuring seamless integration into diverse workflows.
Abeer Badawi, Md Tahmid Rahman Laskar, Jimmy Xiangji Huang
et al.
This position paper argues for a fundamental shift in how Large Language Models (LLMs) are integrated into the mental health care domain. We advocate for their role as co-creators rather than mere assistive tools. While LLMs have the potential to enhance accessibility, personalization, and crisis intervention, their adoption remains limited due to concerns about bias, evaluation, over-reliance, dehumanization, and regulatory uncertainties. To address these challenges, we propose two structured pathways: SAFE-i (Supportive, Adaptive, Fair, and Ethical Implementation) Guidelines for ethical and responsible deployment, and HAAS-e (Human-AI Alignment and Safety Evaluation) Framework for multidimensional, human-centered assessment. SAFE-i provides a blueprint for data governance, adaptive model engineering, and real-world integration, ensuring LLMs align with clinical and ethical standards. HAAS-e introduces evaluation metrics that go beyond technical accuracy to measure trustworthiness, empathy, cultural sensitivity, and actionability. We call for the adoption of these structured approaches to establish a responsible and scalable model for LLM-driven mental health support, ensuring that AI complements, rather than replaces, human expertise.
In this work, we present two novel contributions toward improving research in human-machine teaming (HMT): 1) a Minecraft testbed to accelerate testing and deployment of collaborative AI agents and 2) a tool to allow users to revisit and analyze behaviors within an HMT episode to facilitate shared mental model development. Our browser-based Minecraft testbed allows for rapid testing of collaborative agents in a continuous-space, real-time, partially-observable environment with real humans without cumbersome setup typical to human-AI interaction user studies. As Minecraft has an extensive player base and a rich ecosystem of pre-built AI agents, we hope this contribution can help to facilitate research quickly in the design of new collaborative agents and in understanding different human factors within HMT. Our mental model alignment tool facilitates user-led post-mission analysis by including video displays of first-person perspectives of the team members (i.e., the human and AI) that can be replayed, and a chat interface that leverages GPT-4 to provide answers to various queries regarding the AI's experiences and model details.
Melissa Nevarez-Brewster, Anna M. Zhou, Jenalee R. Doom
et al.
Background: Sleep problems and depression symptoms are highly prevalent and dynamic during pregnancy with impacts on both maternal and offspring health. However, few studies have examined their bidirectional relations across pregnancy to determine whether sleep is an independent predictor of later depression symptoms, and vice versa. The purpose of this study is to investigate the dynamic relations between prenatal maternal sleep problems and depression symptoms three times across pregnancy. Method: Pregnant participants (n = 222) recruited in early pregnancy completed sleep and depression questionnaires at around 17-, 28-, and 35-weeks’ gestation. Prenatal maternal sleep quality was assessed using the Pittsburgh Sleep Quality Index, while depression symptoms were assessed using the Edinburgh Postnatal Depression Scale. Cross-lagged panel models were utilized to examine autoregressive and cross-lagged associations between sleep problems and depression symptoms across pregnancy. Results: Findings reveal that both sleep problems and depression symptoms independently predict one another across pregnancy. All associations covaried for baseline sleep problems and depression symptoms. Specifically, more sleep problems in early pregnancy predicted higher depression symptoms mid-pregnancy (β=.14, p = .011), and elevated depression symptoms in early pregnancy predicted more sleep problems mid-pregnancy (β=.18, p = .002). Similarly, more sleep problems in mid-pregnancy predicted more depression symptoms in late pregnancy (β =.10, p = .029), while depression symptoms in mid-pregnancy predicted more sleep problems during late pregnancy (β=.16, p = .004). Conclusion: Both prenatal maternal sleep and depression independently predict one another across pregnancy. Assessing sleep in early pregnancy may help with the detection of worsening depression, and vice versa, across the perinatal period.
Elisabeth A. Arens, Johannes Falck, Igor Nenadic
et al.
There is evidence that depressed patients differ from non-depressed controls in their fear of happiness. This raises the question of whether fear of happiness causally impact depressive symptoms. The present study investigates the causal impact of fear of happiness on current depressed mood in depressed patients (n = 40) and nondepressed controls (n = 60) by experimentally manipulating fear of happiness. Level of depressed mood state was assessed before and after the manipulation. Depressed patients scored significantly higher on fear of happiness than nondepressed controls. Compared to a control group, participants who were experimentally induced to fear happiness, subsequently exhibited higher levels of depressed mood. This finding was evident in both groups, depressed patients and nondepressed controls. While further research is needed, the present results suggest the importance of fostering positive conceptions of happiness in the treatment of clinical depression as well as in its prevention.
This paper is an ethnographic description of cannabis use in a Canadian Inuit settlement. Cannabis is pervasive in Inuit communities, and both Inuit and public health authorities see it as a serious health and social problem. There is a general understanding that Inuit smoke cannabis to cope with stressors, but little investigation why Inuit choose cannabis instead of other options for managing stress. We describe how cannabis is a pathway through which cash and information circulate in communities, focusing on how smoking cannabis socially provides culturally appropriate forms of support for men experiencing stress. The interpersonal interactions occurring in the context of social cannabis use suggest persistence of traditional healing practices despite historical suppression, revealing how Inuit understandings of personhood and stress continue to shape how men offer support to one another. The positive social and psychological functions of cannabis, however, exist alongside the negative health effects and the economic and interpersonal consequences of excessive use.
The COVID-19 pandemic has shown that all emergencies, major incidents and disease outbreaks can have substantial mental health consequences, and it has demonstrated the proven need for additional care for populations in the wake of disasters. This book brings together practice and recent developments in pre-hospital emergency care, emergency medicine and major trauma care with the wellbeing, psychosocial and mental health aspects of preparing for and responding to emergencies, incidents, terrorism, disasters, epidemics, and pandemics. Practical suggestions are included for future planning to provide better care for people caught up in emergencies. Setting it apart from other books on emergency preparedness is its specific focus on the psychosocial demands imposed on staff of healthcare and responding services. Featuring expert contributions from a wide variety of disciplines, this book appeals to people working within mental healthcare, emergency care, pre-hospital medicine, Blue Light services, public health, humanitarian care, emergency planning, and disaster management.
Mental health is fundamental to an individual's general well-being and ability to work effectively. A trial carried out on college/university students in Ganzhou, Jiangxi Province showed that the mental health of college students could be improved by art healing. The changes in the scores of SCL-90 before and after the trial can be seen that the experimental group had a significant improvement in total score from pre-intervention to post-intervention. For the specific factor items, the anxiety, hostility and additional items were more significant improvements, however, phobia anxiety, paranoid ideation and psychoticism did not show significant differences between pre-intervention and post-intervention. Overall, close to half of the students showed a significant difference in their mental health compared to the pre-trial, and this treatment effect was more pronounced for students with mental health issues. Female students, on the other hand, demonstrated a more positive healing effect than male students. The students in the experimental group showed a reduction in positive psychological symptoms after the art therapy intervention, but did not yet show “effective” results in terms of grade reduction rates. Therefore, a long-term, larger-scale, and more comprehensively designed follow-up trial is necessary to improve the mental health of college students.
Anita Beigzadeh, Vahid Yazdnian, Kamaledin Setarehdan
Any person in his/her daily life activities experiences different kinds and various amounts of mental stress which has a destructive effect on their performance. Therefore, it is crucial to come up with a systematic way of stress management and performance enhancement. This paper presents a comprehensive portable and real-time biofeedback system that aims at boosting stress management and consequently performance enhancement. For this purpose, a real-time brain signal acquisition device, a wireless vibration biofeedback device, and a software-defined program for stress level classification have been developed. More importantly, the entire system has been designed to present minimum time delay by propitiously bridging all the essential parts of the system together. We have presented different signal processing and feature extraction techniques for an online stress detection application. Accordingly, by testing the stress classification section of the system, an accuracy of 83% and a recall detecting the true mental stress level of 92% was achieved. Moreover, the biofeedback system as integrity has been tested on 20 participants in the controlled experimental setup. Experiment evaluations show promising results of system performances, and the findings reveal that our system is able to help the participants reduce their stress level by 55% and increase their accuracy by 24.5%. It can be concluded from the observations that all primary premises on stress management and performance enhancement through reward learning are valid as well.
Sakineh Gerayllo, Najmeh Shahini, Somayeh Ghorbani
et al.
Background: Aging and increase in the elderly population are the most important issues in recent years in developed countries and some developing countries. Mental health is one of the crucial aspects of elderly well-being and must be addressed to improve community health. This study is conducted to examine the mental health of the elderly and the related factors in Gorgan city, 2022.
Methods: The present study is descriptive-analytical and was conducted using a two-stage cluster random sampling method. First, four comprehensive health service centers were selected as a cluster from among the centers of Gorgan city, and then 263 people were randomly selected from the chosen centers, proportionate to the number of registered elderly individuals in the electronic service system. The standard depression questionnaire for the elderly was utilized for assessment. Data were then entered into SPSS software version 18 and with descriptive and analytical statistical tests, including the Mann-Whitney test, Chi-square test, and logistic regression (The significance level was less than 0.05(.
Results: A total of 263 elderly people were evaluated with a mean age of
68.64 ± 7.2 . 143 (54.4%) were male and 120 (45.6%) were female. The prevalence of depression was reported to be 27.8. The occurrence of depression was significantly associated with age, marital status, and the presence of underlying disease conditions (p < 0.05). However, gender, place of residence, and BMI did not show a statistically significant relationship with the occurrence of depression. Aging, underlying diseases, and not being married increase the chance of depression in the elderly.
Conclusion: Although the prevalence of depression in the elderly is lower than the average of the whole country in this study, they are in a better condition. However, due to the fact that they are a special target group, as well as the various risk factors of the disease, including age, lack of marriage, and the presence of an underlying disease in this group, it is suggested that more appropriate planning be done to improve their mental health status
Communities. Classes. Races, Social pathology. Social and public welfare. Criminology