Background: Recent suicide rates among Black American emerging adults have increased sharply, outpacing other racial groups. Stigma plays a critical role in shaping risk for suicidal thoughts and behaviors; however, little research that has examined the ways in which perceptions of perceived public stigma and self-stigma/prejudice are related to suicidal thoughts and behaviors among Black American emerging adults. Methods: A sample of 1224 Black American emerging adults from the general population aged 18-29 were recruited through Qualtrics Panels. Suicidal ideation and attempts were assessed using adapted items from the World Health Organization's suicide screen. Both public and self-stigma/prejudice were evaluated using modified versions of the Devaluation and Discrimination Scale. We used modified Poisson regression, adjusting for race/ethnicity, age, gender, education, health insurance, and treatment utilization. Results: Among those who perceived needing help for mental health concerns, higher perceived public stigma was associated with greater 12-month suicidal ideation, whereas self-stigma was associated with 12-month suicide attempts. These associations were attenuated or absent among those who did not perceive need, except for a pronounced association between prejudice and suicide attempts in sensitivity models. Discussion: Our findings underscore the importance of addressing both perceived public stigma alongside self-stigma/prejudice in prevention efforts to reduce suicide risk in Black emerging adults. Future research is needed to explore mechanisms underlying these associations and to develop targeted prevention strategies.
Large language models (LLMs) are increasingly applied in mental health support systems, where reliable recognition of high-risk states such as suicidal ideation and self-harm is safety-critical. However, existing evaluations primarily rely on aggregate performance metrics, which often obscure risk-specific failure modes and provide limited insight into model behavior in realistic, multi-turn interactions. We present MHDash, an open-source platform designed to support the development, evaluation, and auditing of AI systems for mental health applications. MHDash integrates data collection, structured annotation, multi-turn dialogue generation, and baseline evaluation into a unified pipeline. The platform supports annotations across multiple dimensions, including Concern Type, Risk Level, and Dialogue Intent, enabling fine-grained and risk-aware analysis. Our results reveal several key findings: (i) simple baselines and advanced LLM APIs exhibit comparable overall accuracy yet diverge significantly on high-risk cases; (ii) some LLMs maintain consistent ordinal severity ranking while failing absolute risk classification, whereas others achieve reasonable aggregate scores but suffer from high false negative rates on severe categories; and (iii) performance gaps are amplified in multi-turn dialogues, where risk signals emerge gradually. These observations demonstrate that conventional benchmarks are insufficient for safety-critical mental health settings. By releasing MHDash as an open platform, we aim to promote reproducible research, transparent evaluation, and safety-aligned development of AI systems for mental health support.
Sontaga G. Forane, Absalom E. Ezugwu, Kevin Igwe
et al.
South Africa's escalating mental health crisis, compounded by limited access to culturally responsive care, calls for innovative and contextually grounded interventions. While large language models show considerable promise for mental health support, their predominantly Western-centric training data limit cultural and linguistic applicability in African contexts. This study introduces a proof-of-concept framework that integrates cognitive behavioral therapy with the African philosophy of Ubuntu to create a culturally sensitive, emotionally intelligent, AI-driven mental health dialogue system. Guided by a design science research methodology, the framework applies both deep theoretical and therapeutic adaptations as well as surface-level linguistic and communicative cultural adaptations. Key CBT techniques, including behavioral activation and cognitive restructuring, were reinterpreted through Ubuntu principles that emphasize communal well-being, spiritual grounding, and interconnectedness. A culturally adapted dataset was developed through iterative processes of language simplification, spiritual contextualization, and Ubuntu-based reframing. The fine-tuned model was evaluated through expert-informed case studies, employing UniEval for conversational quality assessment alongside additional measures of CBT reliability and cultural linguistic alignment. Results demonstrate that the model effectively engages in empathetic, context-aware dialogue aligned with both therapeutic and cultural objectives. Although real-time end-user testing has not yet been conducted, the model underwent rigorous review and supervision by domain specialist clinical psychologists. The findings highlight the potential of culturally embedded emotional intelligence to enhance the contextual relevance, inclusivity, and effectiveness of AI-driven mental health interventions across African settings.
Glenna Nightingale, Karthik Mohan, Eloi Ribe
et al.
Background: The negative effects of the COVID-19 pandemic on the mental health and well-being of populations are an important public health issue. Our study aims to determine the underlying factors shaping mental health trajectories during the COVID-19 pandemic in the UK. Methods: Data from the Understanding Society COVID-19 Study were utilized and the core analysis focussed on GHQ36 scores as the outcome variable. We used GAMs to evaluate trends over time and the role of sociodemographic variables, i.e., age, sex, ethnicity, country of residence (in UK), job status (employment), household income, living with a partner, living with children under age 16, and living with a long-term illness, on the variation of mental health during the study period. Results: Statistically significant differences in mental health were observed for age, sex,ethnicity, country of residence (in UK), job status (employment), household income, living with a partner, living with children under age 16, and living with a long-term illness. Women experienced higher GHQ36 scores relative to men with the GHQ36 score expected to increase by 1.260 (95%CI: 1.176, 1.345). Individuals living without a partner were expected to have higher GHQ36 scores, of 1.050 (95%CI: 0.949, 1.148) more than those living with a partner, and age groups 16-34, 35-44, 45-54, 55-64 experienced higher GHQ36 scores relative to those who were 65+. Individuals with relatively lower household income were likely to have poorer mental health relative to those who were more well off. Conclusion: This study identifies key demographic determinants shaping mental health trajectories during the COVID-19 pandemic in the UK. Policies aiming to reduce mental health inequalities should target women, youth, individuals living without a partner, individuals living with children under 16, individuals with a long-term illness, and lower income families.
Violence and rape against women are severe and persistent issues in India, revealing deep-rooted social and cultural challenges. Despite existing legal frameworks and societal efforts, these forms of violence continue to be widespread and inadequately addressed. This article investigates how religious teachings, particularly those from Hinduism and Islam, can offer viable solutions to these problems. By analysing religious texts, the study explores how religious principles can be employed to combat domestic violence and sexual assault, emphasising teachings that advocate for respect, dignity and justice for women. The article provides practical recommendations for integrating faith-based approaches into legal and social strategies, suggesting that incorporating religious values into public discourse and policy can enhance responses to violence and contribute to a more just and safe society. In this context, the article employs a qualitative and textual analysis approach to explore the potential of religious teachings in addressing violence and rape against women in India. Additionally, the study incorporates a quantitative research methodology involving collecting and analysing numerical data to identify patterns, trends and relationships. This combined methodology enables a comprehensive examination of how religious teachings can be utilised to offer ethical and culturally resonant solutions to the critical issues of violence and rape against women in India.
Fabio Rapisarda, Concetta Mezzatesta, Antonina Butticè
et al.
Background and Objectives: Systemic Lupus Erythematosus (SLE) is a chronic autoimmune inflammatory disease affecting connective tissues, with the potential to impact various organs and systems, thereby limiting the quality of life for affected individuals given that it influences their psychological well-being. Indeed, various stress-inducing factors can lead to alterations in emotional regulation, often presenting, as difficulties in identifying and describing emotions, such as those associated with alexithymia. The aim of the present study is to investigate a potential correlation between the incidence of alexithymia and the worsening of symptoms, subsequently influencing the quality of life in individuals with SLE. Materials and Methods: For the realisation of the study a sample of 53 Systemic Lupus Erythematosus (SLE) patients, 47 females and 6 males, aged 16–59 years, was examined and compared with a control group (55 healthy subjects, aged 20–65). The group of patients was recruited within the Connetiviti outpatients’ clinic of the UOC (Complex Oparational Unit) of Rheumatology of the Paolo Giaccone University Polyclinic in Palermo, consisting of healthy subjects and/or with known diagnoses identified within non-clinical contexts, chosen randomly (statistical sense) in order to better represent the general population of the territorial context.Specific psycho-diagnostic measures were administered: Toronto Alexithymia Scale (TAS-20), Response Evaluation Measure (REM-71),Attachment Style Questionnaire (ASQ), Eysenck Personality Inventory (EPI),Symptom Check list (SCL-90), Short Form (SF-36). The selection of these instruments allowed us to assess attachment styles, the presence of maladaptive personality traits, defence mechanisms employed, in order to define any symptoms indicative of a psychiatric diagnosis, and evaluate the quality of life in relation to the severity of the pathology. Results: Comparisons between the two groups revealed significant differences in the Systemic Lupus Erythematosus (SLE) patients, compared to the control group, about the difficulty in identifying feelings and distinguishing them from the bodily sensations accompanying emotional activation, the use of conversion as a defence mechanism and the production of positive symptoms.Moreover, in the same group, the difficulty in identifying emotions was correlated with quality of life, since those who have difficulty identifying emotions have a poorer quality of life.Further analysis regarding the construct of alexithymia, within the patient group alone, revealed a tendency towards somatisation and the use of immature defence mechanisms, including conversion, acting out, projection, dissociation and displacement, highlighting a tendency to act out and project emotional suffering.Furthermore, there appeared to be difficulty in sharing emotional distress with others, indicating a deep sense of insecurity and a need for approval, as well as an incidence in the perception of the state of illness leading to a decline in the quality of life. Conclusions: The study demonstrated that difficulties relating to the identification and processing of emotions impact the disease presentation, influencing the worsening of symptoms and, overall, compromising the patients' quality of life.
The emergence of Small Language Models (SLMs) as privacy-preserving alternatives for sensitive applications raises a fundamental question about their inherent understanding capabilities compared to Large Language Models (LLMs). This paper investigates the mental health understanding capabilities of current SLMs through systematic evaluation across diverse classification tasks. Employing zero-shot and few-shot learning paradigms, we benchmark their performance against established LLM baselines to elucidate their relative strengths and limitations in this critical domain. We assess five state-of-the-art SLMs (Phi-3, Phi-3.5, Qwen2.5, Llama-3.2, Gemma2) against three LLMs (GPT-4, FLAN-T5-XXL, Alpaca-7B) on six mental health understanding tasks. Our findings reveal that SLMs achieve mean performance within 2\% of LLMs on binary classification tasks (F1 scores of 0.64 vs 0.66 in zero-shot settings), demonstrating notable competence despite orders of magnitude fewer parameters. Both model categories experience similar degradation on multi-class severity tasks (a drop of over 30\%), suggesting that nuanced clinical understanding challenges transcend model scale. Few-shot prompting provides substantial improvements for SLMs (up to 14.6\%), while LLM gains are more variable. Our work highlights the potential of SLMs in mental health understanding, showing they can be effective privacy-preserving tools for analyzing sensitive online text data. In particular, their ability to quickly adapt and specialize with minimal data through few-shot learning positions them as promising candidates for scalable mental health screening tools.
Large language models (LLMs) hold significant potential for mental health support, capable of generating empathetic responses and simulating therapeutic conversations. However, existing LLM-based approaches often lack the clinical grounding necessary for real-world psychological counseling, particularly in explicit diagnostic reasoning aligned with standards like the DSM/ICD and incorporating diverse therapeutic modalities beyond basic empathy or single strategies. To address these critical limitations, we propose PsyLLM, the first large language model designed to systematically integrate both diagnostic and therapeutic reasoning for mental health counseling. To develop PsyLLM, we design a novel automated data synthesis pipeline that processes real-world mental health posts collected from Reddit, where users frequently share psychological distress and seek community support. This pipeline processes real-world mental health posts, generates multi-turn dialogue structures, and leverages LLMs guided by international diagnostic standards (e.g., DSM/ICD) and multiple therapeutic frameworks (e.g., CBT, ACT, psychodynamic) to simulate detailed clinical reasoning processes. Rigorous multi-dimensional filtering ensures the generation of high-quality, clinically aligned dialogue data. In addition, we introduce a new benchmark and evaluation protocol, assessing counseling quality across four key dimensions. Our experiments demonstrate that PsyLLM significantly outperforms state-of-the-art baseline models on this benchmark. The model weights and dataset have been publicly released at https://github.com/Emo-gml/PsyLLM.
We present DASH (Deception-Augmented Shared mental model for Human-machine teaming), a novel framework that enhances mission resilience by embedding proactive deception into Shared Mental Models (SMM). Designed for mission-critical applications such as surveillance and rescue, DASH introduces "bait tasks" to detect insider threats, e.g., compromised Unmanned Ground Vehicles (UGVs), AI agents, or human analysts, before they degrade team performance. Upon detection, tailored recovery mechanisms are activated, including UGV system reinstallation, AI model retraining, or human analyst replacement. In contrast to existing SMM approaches that neglect insider risks, DASH improves both coordination and security. Empirical evaluations across four schemes (DASH, SMM-only, no-SMM, and baseline) show that DASH sustains approximately 80% mission success under high attack rates, eight times higher than the baseline. This work contributes a practical human-AI teaming framework grounded in shared mental models, a deception-based strategy for insider threat detection, and empirical evidence of enhanced robustness under adversarial conditions. DASH establishes a foundation for secure, adaptive human-machine teaming in contested environments.
The processes of health building design issues overlap, like the complexity of architecture, technology, and protection of human well-being. It becomes necessary to use a holistic and empathized approach. They meet the concept of the New European Bauhaus (NEB) in terms of attention to the aspects of comprehensive design with a focus on humans and their environment. The investigation focused on psychiatric hospitals with an ever-growing demand for treatment places. Accordingly, this article shows the healing architecture’s examination and the environment in healthcare facilities. The POE method was used by investigating the examples. Research contained the technical, functional, spatial, and behavioral qualities of existing psychiatric hospitals. By presenting elements that positively affect the well-being of users, we indicate good practices that bring psycho-physical benefits.
The demand of qualified nurses is growing worldwide which has made nursing a highly sought-after profession. This profession is revered for its unwavering dedication to healing and support, and stands as a beacon of compassion and care. Yet, there are various challenges that often go unnoticed behind the scenes of this noble vocation, the most disregarded is the mental health struggles faced by nurses. Despite this profession being auspicious, its burden can take a toll on the mental well-being of nurses. However, it is seen as stigma that the people who heal others also need healing. Nurses navigate the complexities of patient care and are the embodiment of resilience and selflessness. On the daily basis, they witness pain, triumph and loss, and shoulder the emotional weight of the patients as well. Moreover, it is often required from them to work irregular hours and rotating shifts, which may include nights, weekends and holidays, which not only disrupts sleep patterns but also makes it challenging for them to maintain work life balance straining personal relationships. These negative aspects can cumulatively result in professional burnout and dissatisfaction in nurses. It may worsen the mental health of nursing leading them to leave the profession altogether. Unfortunately, this issue is not appropriately addressed making it a pervasive stigma in the profession. Nurses are often seen as strong and resilient but it creates a perception that nurses should be able to handle any mental or emotional stress themselves. If the nurses admit vulnerability and seek help, it is viewed as sign of weakness. In addition, nurses fear judgement and the negative opinions of the colleagues who perceive them as incompetent, unreliable, or unfit for duty. Moreover, unfavorable health care settings, such as understaffing, high workload, and limited resources may contribute to the mental health support being viewed as a luxury and inconvenience rather than a legitimate need. These notions hinder the open dialogue and impede access to much needed support. It is imperative that the stakeholders within the healthcare system exert efforts to address the challenges faced by nurses regarding mental. We need to acknowledge and recognize that the arduousness of the nursing profession and how it affects the mental well-being of nurses. The organizations should prioritize mental health of their practitioners by offering them resources such as counseling services and peer support networks. The stigma surrounding mental health needs to be shattered so that an environment can be fostered where nurses feel safe and are supported in their time of vulnerability. Only then can we ensure that auspicious nature of nursing shines brighter than ever before and illuminate a path of healing for healers as well.
Substance Use Disorder (SUD) is associated with harmful outcomes and contributes to antisocial behavior and psychological disorders. The present study evaluated the effectiveness of Group Metacognitive Therapy (g-MCT) on psychological symptoms in Muslim women with SUD in the Herat Province of Afghanistan. In this study, Muslim women with substance use disorder (N = 30, mean age: 38.15 years, 100% female, 100% Afghan) were randomized to Metacognitive Therapy (MCT) or Treatment as Usual (TAU). The MCT group received six sessions of MCT, and the TAU group received only methadone maintenance therapy. All participants completed a demographic questionnaire, Metacognition Questionnaire, Beck Anxiety Inventory, Beck Depression Inventory-II, Drug Use Evaluation Questionnaire, Personal Concerns Inventory, Motivational Structure Questionnaire, and Situational Confidence Questionnaire (SCQ). Data was collected at baseline and the 3-month follow-up sessions. Compared with TAU, the MCT group was effective in reducing anxiety, depression, substance use evaluation, and personal concerns, as well as improving motivational structure, situational confidence, and metacognitive beliefs in Afghani women with Substance Use Disorder. MCT added to methadone maintenance therapy was acceptable and improved outcomes. This study suggests that MCT is an effective treatment for improving psychological symptoms in Muslim women with SUD, but only in mild to moderate severity cases. For people with severe Substance Use Disorder, additional treatment may be necessary.
Nafisa Ferdous, María Luisa Zúñiga, Kelly E. Courtney
Abstract The influence of alcohol use on later neurocognitive functioning is well researched, yet few studies have investigated whether neurocognition post‐drinking initiation in adolescence predicts changes in later alcohol use. The objective of this study was to investigate neurocognitive task performance during maximum alcohol use in late adolescence as predictors of drinking behaviors 3–7 years later. Analyses (n = 105) were conducted on a longitudinal data set involving adolescents (12–13 years old) who were followed for 16 years. Time 1 (T1) was defined as the individuals' maximum drinking year within the first 10 study years and Time 2 (T2) was the first available data entry 3–7 years after T1. Four hierarchical linear regression models predicting follow‐up alcohol use were estimated: drinking days, average drinks per drinking day, peak drinks, and binge episodes. All models included inhibition/cognitive flexibility, visuospatial ability, verbal memory, working memory, and their interactions with sex, while covarying for age at T1, follow‐up duration, and controlling for T1 drinking. Better visuospatial ability at T1 predicted decreases in later binge episodes at T2 (β = −0.19, p = 0.048, partial r2 = 0.039). While better inhibition/cognitive flexibility at T1 predicted increases in follow‐up drinks per drinking day at T2 (β = 0.18, p = 0.016, partial r2 = 0.057). Findings suggest specific neurocognitive abilities during maximum drinking in late adolescence are useful as predictors of change in later drinking quantity per occasion and could potentially inform intervention research targeting this age group.
Jill C. Fodstad, Lauren B. Jones, Micah Iticovici
et al.
Adolescents and adults with Down syndrome are noted to display symptoms consistent with various anxiety disorders. While evidenced-based practices, including psychotherapies and psychopharmacology, exist and effectively treat anxiety in neurotypical populations, less is known about anxiety treatments for persons with Down syndrome. A scoping rapid review was conducted in April 2023 to determine what treatments are being used to target anxiety in adolescents and adults with Down syndrome, the quality of those treatments, and their alignment with current evidence-based practices. A total of eleven articles, primarily single case or case series, published between 1981 and 2022 were identified targeting adolescents and adults with Down syndrome diagnosed with specific phobias, selective mutism, generalized anxiety disorder, agoraphobia with panic, and non-specific anxiety symptoms. Interventions used most often aligned with evidence-based anxiety treatment guidelines and included psychotherapy, complementary and alternative medicine, and psychopharmacology. While most studies reported positive treatment responses showing reductions in anxiety symptoms post-treatment, the quality and generalizability of the studies was primarily poor. More rigorous research evaluating the effects of treatment for anxiety symptoms in the DS population are needed to develop guidelines to address anxiety disorders in this vulnerable population.
Large language models (LLMs) are already being piloted for clinical use in hospital systems like NYU Langone, Dana-Farber and the NHS. A proposed deployment use case is psychotherapy, where a LLM-powered chatbot can treat a patient undergoing a mental health crisis. Deployment of LLMs for mental health response could hypothetically broaden access to psychotherapy and provide new possibilities for personalizing care. However, recent high-profile failures, like damaging dieting advice offered by the Tessa chatbot to patients with eating disorders, have led to doubt about their reliability in high-stakes and safety-critical settings. In this work, we develop an evaluation framework for determining whether LLM response is a viable and ethical path forward for the automation of mental health treatment. Our framework measures equity in empathy and adherence of LLM responses to motivational interviewing theory. Using human evaluation with trained clinicians and automatic quality-of-care metrics grounded in psychology research, we compare the responses provided by peer-to-peer responders to those provided by a state-of-the-art LLM. We show that LLMs like GPT-4 use implicit and explicit cues to infer patient demographics like race. We then show that there are statistically significant discrepancies between patient subgroups: Responses to Black posters consistently have lower empathy than for any other demographic group (2%-13% lower than the control group). Promisingly, we do find that the manner in which responses are generated significantly impacts the quality of the response. We conclude by proposing safety guidelines for the potential deployment of LLMs for mental health response.
Onno P. Kampman, Ye Sheng Phang, Stanley Han
et al.
We introduce a general-purpose, human-in-the-loop dual dialogue system to support mental health care professionals. The system, co-designed with care providers, is conceptualized to assist them in interacting with care seekers rather than functioning as a fully automated dialogue system solution. The AI assistant within the system reduces the cognitive load of mental health care providers by proposing responses, analyzing conversations to extract pertinent themes, summarizing dialogues, and recommending localized relevant content and internet-based cognitive behavioral therapy exercises. These functionalities are achieved through a multi-agent system design, where each specialized, supportive agent is characterized by a large language model. In evaluating the multi-agent system, we focused specifically on the proposal of responses to emotionally distressed care seekers. We found that the proposed responses matched a reasonable human quality in demonstrating empathy, showing its appropriateness for augmenting the work of mental health care providers.
LGBTQ+ individuals are increasingly turning to chatbots powered by large language models (LLMs) to meet their mental health needs. However, little research has explored whether these chatbots can adequately and safely provide tailored support for this demographic. We interviewed 18 LGBTQ+ and 13 non-LGBTQ+ participants about their experiences with LLM-based chatbots for mental health needs. LGBTQ+ participants relied on these chatbots for mental health support, likely due to an absence of support in real life. Notably, while LLMs offer prompt support, they frequently fall short in grasping the nuances of LGBTQ-specific challenges. Although fine-tuning LLMs to address LGBTQ+ needs can be a step in the right direction, it isn't the panacea. The deeper issue is entrenched in societal discrimination. Consequently, we call on future researchers and designers to look beyond mere technical refinements and advocate for holistic strategies that confront and counteract the societal biases burdening the LGBTQ+ community.
Verbal communication plays a crucial role in human cooperation, particularly when the partners only have incomplete information about the task, environment, and each other's mental state. In this paper, we propose a novel cooperative communication framework, Goal-Oriented Mental Alignment (GOMA). GOMA formulates verbal communication as a planning problem that minimizes the misalignment between the parts of agents' mental states that are relevant to the goals. This approach enables an embodied assistant to reason about when and how to proactively initialize communication with humans verbally using natural language to help achieve better cooperation. We evaluate our approach against strong baselines in two challenging environments, Overcooked (a multiplayer game) and VirtualHome (a household simulator). Our experimental results demonstrate that large language models struggle with generating meaningful communication that is grounded in the social and physical context. In contrast, our approach can successfully generate concise verbal communication for the embodied assistant to effectively boost the performance of the cooperation as well as human users' perception of the assistant.
In this paper, we define a novel recursive Heaviside step sequence function and demonstrate its applicability to modeling human mental states such as thought processes, memory recall, and forgetfulness. By extending the traditional Heaviside step function, which typically represents binary transitions, into a recursive sequence framework, we introduce a dynamic model that better captures the complexities of cognitive states. Furthermore, the recursive Heaviside step sequence function approximates solutions to a multidimensional inviscid advection equation, offering a unique mathematical perspective on the evolution of mental states over time. This continuous model, combined with the recursive delta sequence function, provides a comprehensive approach to exploring how memories and thoughts emerge, evolve, and fade. Through this approach, we propose that mental states can be expressed as time series functions, and the selection of the parameter N reflects individual variability in mental processing, influenced by external environments and internal experiences. We also discuss the implications of this framework for understanding human cognition and potential limitations due to modern technological constraints in replicating such processes.