Suraj Racha, Prashant Harish Joshi, Utkarsh Maurya
et al.
Large Language Models (LLMs) have shown remarkable capabilities for complex tasks, yet adaptation in medical domain, specifically mental health, poses specific challenges. Mental health is a rising concern globally with LLMs having large potential to help address the same. We highlight three primary challenges for LLMs in mental health - lack of high quality interpretable and knowledge grounded training data; training paradigms restricted to core capabilities, and evaluation of multi turn dialogue settings. Addressing it, we present oMind framework which includes training and aligning LLM agents for diverse capabilities including conversations; high quality ~164k multi-task SFT dataset, as a result of our generation pipeline based on Structured Knowledge retrieval, LLM based pruning, and review actions. We also introduce oMind-Chat - a novel multi turn benchmark dataset with expert annotated turn level and conversation level rubrics. Our diverse experiments on both core capabilities and conversations shows oMind LLMs consistently outperform baselines. oMind-LLM also shows significantly better reasoning with up to 80% win rate.
Mental manipulation, the strategic use of language to covertly influence or exploit others, is a newly emerging task in computational social reasoning. Prior work has focused exclusively on textual conversations, overlooking how manipulative tactics manifest in speech. We present the first study of mental manipulation detection in spoken dialogues, introducing a synthetic multi-speaker benchmark SPEECHMENTALMANIP that augments a text-based dataset with high-quality, voice-consistent Text-to-Speech rendered audio. Using few-shot large audio-language models and human annotation, we evaluate how modality affects detection accuracy and perception. Our results reveal that models exhibit high specificity but markedly lower recall on speech compared to text, suggesting sensitivity to missing acoustic or prosodic cues in training. Human raters show similar uncertainty in the audio setting, underscoring the inherent ambiguity of manipulative speech. Together, these findings highlight the need for modality-aware evaluation and safety alignment in multimodal dialogue systems.
This chapter, Healing Waters: Aquatherapy for Mental Resilience and Injury Prevention, delves into the transformative therapeutic benefits of hydrotherapy in enhancing both physical and mental well-being. The unique properties of water—buoyancy, resistance, and temperature modulation—create an environment conducive to low impact and efficient movement making it ideal for injury prevention and rehabilitation. Hydrotherapy plays a pivotal role in alleviating physical strain, promoting joint health, enhancing muscular strength and flexibility without the stress associated with weight-bearing exercises. Beyond its physical advantages, water-based activities are shown to have profound psychological benefits. Hydrotherapy reduces stress, elevates mood, and fosters mental resilience by stimulating the parasympathetic nervous system, which promotes relaxation and emotional balance. The chapter highlights the strong interplay between mental health and physical recovery, showcasing how aquatic environments serve as a bridge to holistic healing. Furthermore, this chapter offers evidence-based insights into designing targeted aquatic exercise programmes tailored to diverse populations, such as athletes aiming to prevent injuries, individuals recovering from orthopaedic conditions, and those seeking enhanced mental health. By integrating practical applications and real-world examples, it underscores the potential of hydrotherapy to revolutionise traditional approaches to fitness, therapy, and mental healthcare, providing a holistic pathway to resilience and recovery.
The rapid advancement of Large Language Models (LLMs), reasoning models, and agentic AI approaches coincides with a growing global mental health crisis, where increasing demand has not translated into adequate access to professional support, particularly for underserved populations. This presents a unique opportunity for AI to complement human-led interventions, offering scalable and context-aware support while preserving human connection in this sensitive domain. We explore various AI applications in peer support, self-help interventions, proactive monitoring, and data-driven insights, using a human-centred approach that ensures AI supports rather than replaces human interaction. However, AI deployment in mental health fields presents challenges such as ethical concerns, transparency, privacy risks, and risks of over-reliance. We propose a hybrid ecosystem where where AI assists but does not replace human providers, emphasising responsible deployment and evaluation. We also present some of our early work and findings in several of these AI applications. Finally, we outline future research directions for refining AI-enhanced interventions while adhering to ethical and culturally sensitive guidelines.
Limited access to mental health care has motivated the use of digital tools and conversational agents powered by large language models (LLMs), yet their quality and reception remain unclear. We present a study comparing therapist-written responses to those generated by ChatGPT, Gemini, and Llama for real patient questions. Text analysis showed that LLMs produced longer, more readable, and lexically richer responses with a more positive tone, while therapist responses were more often written in the first person. In a survey with 150 users and 23 licensed therapists, participants rated LLM responses as clearer, more respectful, and more supportive than therapist-written answers. Yet, both groups of participants expressed a stronger preference for human therapist support. These findings highlight the promise and limitations of LLMs in mental health, underscoring the need for designs that balance their communicative strengths with concerns of trust, privacy, and accountability.
Black men face a double barrier to mental health help-seeking: traditional masculinity norms demanding emotional restrictiveness and systemic racism fostering institutional mistrust. While celebrity mental health disclosures show promise for stigma reduction, limited research examines their impact on Black masculine communities through digital platforms. This convergent mixed-methods study analysed 11,306 YouTube comments following rapper Lil Wayne's unprecedented disclosure of childhood suicide attempt and lifelong mental health struggles. Quantitative analysis using VADER sentiment classification, Latent Dirichlet Allocation topic modelling, and NRC emotion lexicon analysis revealed predominantly positive sentiment with systematic community amplification of mental health discourse. Reflexive thematic analysis of 2,100 high-engagement comments identified eight themes, with peer support achieving the highest saturation, contradicting isolation narratives. Findings support a Digital Permission Structures Model demonstrating how intersectional celebrity status (race + gender + high-status), hip-hop authenticity values, and digital platform affordances create triadic authorisation mechanisms enabling vulnerability expression. Community responses revealed communal masculinity rooted in Ubuntu philosophy and active reconstruction of masculine norms, positioning help-seeking as strength. Results challenge deficit-based models of Black masculinity, suggesting interventions should leverage collectivism, partner with high-status cultural figures, employ strength-based messaging, and centre hip-hop authenticity rather than imposing Western individualistic frameworks. This study provides evidence-based strategies for culturally responsive mental health interventions addressing persistent disparities in Black men's service utilisation.
As social media platforms are increasingly adopted, the data the data people leave behind is shining new light into our understanding of phenomena, ranging from socio-economic-political events to the spread of infectious diseases. This chapter presents research conducted in the past decade that has harnessed social media data in the service of mental health and well-being. The discussion is organized along three thrusts: a first that highlights how social media data has been utilized to detect and predict risk to varied mental health concerns; a second thrust that focuses on translation paradigms that can enable to use of such social media based algorithms in the real-world; and the final thrust that brings to the fore the ethical considerations and challenges that engender the conduct of this research as well as its translation. The chapter concludes by noting open questions and problems in this emergent area, emphasizing the need for deeper interdisciplinary collaborations and participatory research design, incorporating and centering on human agency, and attention to societal inequities and harms that may result from or be exacerbated in this line of computational social science research.
This study presents a multi-stage approach to mental health classification by leveraging traditional machine learning algorithms, deep learning architectures, and transformer-based models. A novel data set was curated and utilized to evaluate the performance of various methods, starting with conventional classifiers and advancing through neural networks. To broaden the architectural scope, recurrent neural networks (RNNs) such as LSTM and GRU were also evaluated to explore their effectiveness in modeling sequential patterns in the data. Subsequently, transformer models such as BERT were fine-tuned to assess the impact of contextual embeddings in this domain. Beyond these baseline evaluations, the core contribution of this study lies in a novel training strategy involving a dual-model architecture composed of a teacher and a student network. Unlike standard distillation techniques, this method does not rely on soft label transfer; instead, it facilitates information flow through both the teacher model's output and its latent representations by modifying the loss function. The experimental results highlight the effectiveness of each modeling stage and demonstrate that the proposed loss function and teacher-student interaction significantly enhance the model's learning capacity in mental health prediction tasks.
Mara Vizzuso, Gianluca Passarelli, Giovanni Cantele
et al.
Trotter errors in digitized quantum dynamics arise from approximating time-ordered evolution under noncommuting Hamiltonian terms with a product formula. In the adiabatic regime, such errors are known to exhibit long-time self-healing [Phys. Rev. Lett. \textbf{131}, 060602 (2023)], where discretization effects are effectively suppressed. Here we show that self-healing persists at finite evolution times once nonadiabatic errors induced by finite-speed ramps are compensated. Using counterdiabatic driving to cancel diabatic transitions and isolate discretization effects, we study both noninteracting and interacting spin models and characterize the finite-time scaling with the Trotter steps and the total evolution time. In the instantaneous eigenbasis of the driven Hamiltonian, the leading digital error maps to an effective harmonic perturbation whose dominant Fourier component yields an analytic upper bound on the finite-time Trotter error and reveals the phase-cancellation mechanism underlying self-healing. Our results establish finite-time self-healing as a generic feature of digitized counterdiabatic protocols, clarify its mechanism beyond the long-time adiabatic limit, and provide practical guidance for high-fidelity state preparation on gate-based quantum processors.
Following the recent release of various Artificial Intelligence (AI) based Conversation Agents (CAs), adolescents are increasingly using CAs for interactive knowledge discovery on sensitive topics, including mental and sexual health topics. Exploring such sensitive topics through online search has been an essential part of adolescent development, and CAs can support their knowledge discovery on such topics through human-like dialogues. Yet, unintended risks have been documented with adolescents' interactions with AI-based CAs, such as being exposed to inappropriate content, false information, and/or being given advice that is detrimental to their mental and physical well-being (e.g., to self-harm). In this position paper, we discuss the current landscape and opportunities for CAs to support adolescents' mental and sexual health knowledge discovery. We also discuss some of the challenges related to ensuring the safety of adolescents when interacting with CAs regarding sexual and mental health topics. We call for a discourse on how to set guardrails for the safe evolution of AI-based CAs for adolescents.
Kamala Devi Kannan, Senthil Kumar Jagatheesaperumal, Rajesh N. V. P. S. Kandala
et al.
For the early identification, diagnosis, and treatment of mental health illnesses, the integration of deep learning (DL) and machine learning (ML) has started playing a significant role. By evaluating complex data from imaging, genetics, and behavioral assessments, these technologies have the potential to significantly improve clinical outcomes. However, they also present unique challenges related to data integration and ethical issues. This survey reviews the development of ML and DL methods for the early diagnosis and treatment of mental health issues. It examines a range of applications, with a particular emphasis on behavioral assessments, genetic and biomarker analysis, and medical imaging for diagnosing diseases like depression, bipolar disorder, and schizophrenia. Predictive modeling for illness progression is further discussed, focusing on the role of risk prediction models and longitudinal studies. Key findings highlight how ML and DL can improve diagnostic accuracy and treatment outcomes while addressing methodological inconsistencies, data integration challenges, and ethical concerns. The study emphasizes the importance of building real-time monitoring systems for individualized treatment, enhancing data fusion techniques, and fostering interdisciplinary collaboration. Future research should focus on overcoming these obstacles to ensure the valuable and ethical application of ML and DL in mental health services.
Background: Psychological resilience, as a potential protective factor, has been linked to enhanced wellbeing and reduced psychopathology. This study investigated the predictors, associations, and prospective effects of resilience to daily hassles in both short-term (across days) and long-term (across months) contexts. Methods: Daily resilience was measured through self-reported abilities to cope with daily hassles on a day-to-day basis for 30 days using Experience Sampling Methods in a sample of 86 outpatients diagnosed with depression. Results: Daily resilience correlated with baseline depression (r = -0.31; p < .001), but not with the five other personal and social resources we investigated. As hypothesized, daily resilience co-varied with daily wellbeing on a day-to-day basis (p = <0.001), was predictive of shifts in wellbeing across days (p = <0.01), and associated with patients’ improvements in depression and dysfunction three months later (B = -0.48 & -0.49; p < .001) and six months later (B = -0.55 & -0.57; p < .001). Limitations: It should be noted that the study's participants were recruited from a larger Randomized Controlled Trial (RCT), and approximately 40% of patients experienced daily hassles. Despite this limitation, the findings suggest that daily resilience may hold promise as a crucial factor to inform intervention strategies and programs aimed at preventing depression relapse. Conclusion: Daily resilience may well be a promising factor to inform intervention strategies and programmes that aim to prevent (relapse of) depression.
Previous works from research and industry have proposed a spatial representation of code in a canvas, arguing that a navigational code space confers developers the freedom to organise elements according to their understanding. By allowing developers to translate logical relatedness into spatial proximity, this code representation could aid in code navigation and comprehension. However, the association between developers' code comprehension and their visuo-spatial mental model of the code is not yet well understood. This mental model is affected on the one hand by the spatial code representation and on the other by the visuo-spatial working memory of developers. We address this knowledge gap by conducting an online experiment with 20 developers following a between-subject design. The control group used a conventional tab-based code visualization, while the experimental group used a code canvas to complete three code comprehension tasks. Furthermore, we measure the participants' visuo-spatial working memory using a Corsi Block test at the end of the tasks. Our results suggest that, overall, neither the spatial representation of code nor the visuo-spatial working memory of developers has a significant impact on comprehension performance. However, we identified significant differences in the time dedicated to different comprehension activities such as navigation, annotation, and UI interactions.
Kaushik Roy, Vedant Khandelwal, Raxit Goswami
et al.
After the pandemic, artificial intelligence (AI) powered support for mental health care has become increasingly important. The breadth and complexity of significant challenges required to provide adequate care involve: (a) Personalized patient understanding, (b) Safety-constrained and medically validated chatbot patient interactions, and (c) Support for continued feedback-based refinements in design using chatbot-patient interactions. We propose Alleviate, a chatbot designed to assist patients suffering from mental health challenges with personalized care and assist clinicians with understanding their patients better. Alleviate draws from an array of publicly available clinically valid mental-health texts and databases, allowing Alleviate to make medically sound and informed decisions. In addition, Alleviate's modular design and explainable decision-making lends itself to robust and continued feedback-based refinements to its design. In this paper, we explain the different modules of Alleviate and submit a short video demonstrating Alleviate's capabilities to help patients and clinicians understand each other better to facilitate optimal care strategies.
Ademola Adeponle, Danielle Groleau, Oye Gureje
et al.
Perinatal depression is a major public health problem that is under-treated in low- and middle-income countries, with negative community attitudes often cited as a major barrier to help-seeking and treatment. In this paper, we investigate help-seeking for perinatal depression, some of whom may have experienced psychosis in the context of perinatal depression, and its cultural shaping in Nigeria. Our approach was informed by cultural constructivist and critical anthropological perspectives to situate perinatal depression in ecosocial context with systematic attention to the social-structural determinants. Help-seeking was controlled by caregivers, but patients and caregivers differed in their definitions of illness onset: caregivers focused on violation of norms for maternal behavior, whereas patients focused on somatic and psychological distress. Help-seeking entailed use of two different reasoning approaches. The default approach was ad-hoc and contingency-based, depending on: 1) the kinds of help immediately available and the fit with experiential knowledge; 2) the meanings of the patient's problem as negotiated in social interactions; 3) concerns about stigma that led to secrecy in seeking help; and 4) the perceived degree of immediate risk to safety. The second approach involved more explicit deliberation and took over in situations of uncertainty including: 1) worsening or unremitting illness despite treatment; 2) harm from treatment; and 3) perceived inadequacy of a treatment to bring about healing. Avoiding mental illness stigmatization was seen as dependent on limiting public awareness of the individual's affliction. The meanings of illness were locally informed and negotiated in ongoing social interactions and practices of social recognition and status negotiation that legitimated illness and influenced help-seeking. Clarifying the social context of mental health problems and identifying cultural and structural risk and protective factors can inform the design of health care systems to improve access to care and the development of culturally appropriate and effective intervention programmes.
Madhumitha Balaji, Kavita Mandhare, Kalyani Nikhare
et al.
Suicidal behaviours among young people in India are a major public health problem. An understanding of the reasons for suicide attempts from survivor perspectives is essential to developing suicide prevention programs for this population, as these can provide valuable insights into concerns that are unique to young people, and direct the focus of such programs towards these specific concerns. Qualitative studies are best suited for eliciting such perspectives, but such studies in India are lacking. We conducted semi-structured interviews with 47 persons between the ages of 15 and 29 who had been admitted to a public hospital in Pune, India, following a suicide attempt. Participants were asked to describe in their own words, what they believed was the reason for the attempt. Data was analysed using inductive thematic analysis and summative content analysis. There were three broad factors that interacted to lead to suicide attempts – background factors (individual and environmental factors that increased vulnerability); psychological distress (emotional and cognitive states that led to suicidal ideation); and intervening factors (factors that facilitated transition from distress to the attempt). The most common pattern was the occurrence of an interpersonal stressor shortly before the attempt, which produced distorted cognitions, and overwhelming emotions – usually of anxiety or anger - with ready access to means and/or impulsivity being the final catalysts. This stressor was generally a trigger associated with long-standing problems involving partners or family members, which had already produced intolerable distress over time. Female participants appeared particularly vulnerable to these stressors, partly due to prevailing socio-cultural norms. Our findings suggest the need for suicide prevention programs in India to foster life skills for young people, engage with high-risk groups (for example, women), and restrict access to harmful substances. Family stakeholders need to be involved in the implementation.
Sydney M. Silverstein, Josef Rivera, Danielle Gainer
et al.
While numerous studies have established relationships between Adverse Childhood Experiences (ACEs) and adult substance use, few qualitative studies have explored the differing ways in which experiences of childhood adversity are emplotted into narratives of drug use and recovery. This paper analyzes qualitative data collected as part of a mixed-methods longitudinal study of people with opioid use disorder. Narratives of adverse childhood experiences emerged unprompted. After coding qualitative data for mention of ACEs, we thematically analyzed coded data using a framework of critical phenomenology and constructed a four-part typology to differentiate the ways that ACEs were emplotted into narratives. Our four sub-types—which we call ‘haunted by trauma’, ‘seeking redemption’, ‘casual mentioners’, and ‘reckoning with inevitability’—did not necessarily cleave along types or number of ACEs, but rather by the manners in which these experiences were conditioned by subsequent life trajectories, and the social, structural, and interpersonal factors that contextualized them. While participants often linked experiences of childhood adversity to adult opioid use, we argue that the differing ways in which individuals understand and process these linkages has implications for both clinical and therapeutic practice. For frameworks like trauma-informed care to be meaningful, we must pay closer attention to these meaningful differences.
Subburaj Alagarsamy, Nalayini Sugirthan, Sangeeta Mehrolia
et al.
The Depression Anxiety Stress Scales-21 (DASS-21) is a well-established scale designed to measure the negative emotional states of depression, anxiety, and stress. DASS-21 has been translated into various languages, and findings conclude that it is psychometrically sound, with good reliability and validity. This study adapts and validates the psychometric properties of DASS-21 in the Tamil language. The instrument was administered to 511 Tamil speaking students ranging between 18 and 35 of age with an average age of 21 years. Results reaffirm that DASS-21 three-factor model shows excellent validity and reliability on the entire sample and groups based on age, gender, and residential area. They also find support in different hierarchical variance measurement models (metric, scalar, strict models). This study concludes that Tamil DASS-21 can be used as a universal measure to map symptoms and screen for depression, anxiety, and stress in any circumstances. Our findings provide roadmap for future research on the Tamil version of DASS-21 with specific focus on its clinical use.
Katherine Cummergen, Laura Hannah, Louise Jopling
et al.
Background: It is unclear whether current outcome measures capture what is important to service users experiencing treatment-resistant depression (TRD). This review aims to understand what outcomes are important to people receiving treatment for TRD, and to ascertain how this is being measured or could be measured to aid values-based commissioning in the implementation of specialist services. Methods: A systematic search was conducted across nine databases: EMBASE, PSychINFO, AMED, EMCARE, PubMed, BNI, HMIC, CINHAL, and Medline. Quantitative and qualitative studies, and non-empirical work were included. No publication date restrictions were set. Included studies were appraised for quality. Results: Twenty-two studies met inclusion for the review, including two opinion pieces. Thematic analysis was used to extract five themes: important outcomes beyond recovery from symptoms; differentiations in perspectives; patient preferences; essential sets of outcome measures; and underdeveloped outcome measures from the patient's perspective. Limitations: The search strategy was partially systematic due to the exploratory nature of the subject and the lack of available research in the field. Studies included collect data on patient perspectives but did not demonstrate co-production throughout the whole research process. Conclusions: Outcomes in persistent depression have been neglected, especially from the patient perspective. The findings from this review make an important contribution to agreeing desirable outcomes for people with TRD by drawing together the literature and highlighting how and why it is necessary to apply certain methods to persistent depression. The report identifies areas where further understanding and research is needed and how to inform current service commissioning practices.
Israel L. Donato Ridgley, Randy A. Freeman, Kevin M. Lynch
In this paper we describe a parameterized family of first-order distributed optimization algorithms that enable a network of agents to collaboratively calculate a decision variable that minimizes the sum of cost functions at each agent. These algorithms are self-healing in that their correctness is guaranteed even if they are initialized randomly, agents drop in or out of the network, local cost functions change, or communication packets are dropped. Our algorithms are the first single-Laplacian methods to exhibit all of these characteristics. We achieve self-healing by sacrificing internal stability, a fundamental trade-off for single-Laplacian methods.