Health Chatbots Research Articles

Symmetric Therapeutic Frameworks and Ethical Dimensions in AI-Based Mental Health Chatbots (2020–2025): A Systematic Review of Design Patterns, Cultural Balance, and Structural Symmetry

Artificial intelligence (AI)-powered mental health chatbots have evolved quickly as scalable means for psychological support, bringing novel solutions through natural language processing (NLP), mobile accessibility, and generative AI. This systematic literature review (SLR), following PRISMA 2020 guidelines, collates evidence from 25 published, peer-reviewed studies between 2020 and 2025 and reviews therapeutic techniques, cultural adaptation, technical design, system assessment, and ethics. Studies were extracted from seven academic databases, screened against specific inclusion criteria, and thematically analyzed. Cognitive behavioral therapy (CBT) was the most common therapeutic model, featured in 15 systems, frequently being used jointly with journaling, mindfulness, and behavioral activation, followed by emotion-based approaches, which were featured in seven systems. Innovative techniques like GPT-based emotional processing, multimodal interaction (e.g., AR/VR), and LSTM-SVM classification models (greater than 94% accuracy) showed increased conversation flexibility but missed long-term clinical validation. Cultural adaptability was varied, and effective localization was seen in systems like XiaoE, okBot, and Luda Lee, while Western-oriented systems had restricted contextual adaptability. Accessibility and inclusivity are still major challenges, especially within low-resource settings, since digital literacy, support for multiple languages, and infrastructure deficits are still challenges. Ethical aspects—data privacy, explainability, and crisis plans—were under-evidenced for most deployments. This review is different from previous ones since it focuses on cultural adaptability, ethics, and hybrid public health incorporation and proposes a comprehensive approach for deploying AI mental health chatbots safely, effectively, and inclusively. Central to this review, symmetry is emphasized as a fundamental idea incorporated into frameworks for cultural adaptation, decision-making processes, and therapeutic structures. In particular, symmetry ensures equal cultural responsiveness, balanced user–chatbot interactions, and ethically aligned AI systems, all of which enhance the efficacy and dependability of mental health services. Recognizing these benefits, the review further underscores the necessity for more rigorous academic research into the development, deployment, and evaluation of mental health chatbots and apps, particularly to address cultural sensitivity, ethical accountability, and long-term clinical outcomes.

Conversational health agents: a personalized large language model-powered agent framework

ObjectiveConversational Health Agents (CHAs) are interactive systems providing healthcare services, such as assistance and diagnosis. Current CHAs, especially those utilizing Large Language Models (LLMs), primarily focus on conversation aspects. However, they offer limited agent capabilities, specifically needing more multistep problem-solving, personalized conversations, and multimodal data analysis. We aim to overcome these limitations.Materials and methodsWe propose openCHA, an open-source LLM-powered framework, designed to enable the development of conversational agents. OpenCHA offers a foundational and structured architecture and codebase, enabling researchers and developers to build and customize their CHA based on the specifics of their intended application. The framework leverages knowledge acquisition, problem-solving capabilities, multilingual, and multimodal conversations, and allows interaction with various AI platforms. We have released the framework as open source for the community on GitHub (https://github.com/Institute4FutureHealth/CHA and https://opencha.com).ResultsWe demonstrated the openCHA’s capability to develop CHAs across multiple health domains using 2 demos and 5 use cases. In diabetic patient management, developed CHA achieved a 92.1% accuracy rate, surpassing GPT4’s 51.8%. In food recommendations, developed CHA outperformed GPT4. The developed CHA excelled as an evaluator for mental health chatbots, recording the lowest Mean Absolute Error at 0.31, compared to competitors like GPT, Misteral, Gemini, and Claude. Additionally, the empathy enabled CHA identified emotional states with 89% accuracy, and in physiological data analysis of heart rate from Photoplethysmography (PPG) signals, the developed CHA achieved an mean absolute error of 2.83, far lower than GPT-4o’s 8.93.DiscussionThe openCHA framework enhances CHAs by enabling features such as explainability, personalization, and reliability through its integration with LLMs and external data sources. The developed CHAs face challenges like latency, token limits, and scalability. Future efforts will focus on improving planning robustness, enhancing accuracy and evaluation methods, and resolving user query ambiguity to further refine the framework’s effectiveness.ConclusionThe diverse demos and use cases of openCHA demonstrate the framework’s capacity to empower the development of a wide range of CHAs for various healthcare tasks.

Chatbots and mental health: a scoping review of reviews

Abstract The use of chatbots for mental health is now frequently discussed as a means to increase access to mental health resources. For this scoping review of reviews on chatbots for mental health, we performed a systematic search of the literature on Scopus, Web of Science, PubMed and Dimensions.ai and identified 14 relevant reviews published in scientific journals which identified publications following a systematic search of two or more databases. Three additional relevant reviews were included following forward and backward reference examination. The 17 included reviews are: eight systematic reviews (three with meta-analysis), seven scoping reviews and two unlabeled reviews. We have summarized the scope of the reviews as well as their findings. Most reviews included at least one study with 15 participants or fewer. Few reviews report a significant proportion of randomized controlled trials and some original publications most of the time report on piloting studies. Overall, the reviews examine the opinions of users of chatbots, the features of chatbots, the outcomes and measures in studies relying on chatbots, the effectiveness of chatbots (particularly the meta-analyses) in alleviating mental disorder symptomatology as well as their potential for the assessment of mental health. The use of machine learning and natural language processing frameworks (Dialogflow, RASA) seems to drive an increasing number of studies of chatbots for mental health, but progress is necessary in several areas affecting interaction dynamics.

Mental Health Professionals’ Views on Artificial Intelligence as an Aide for Children Anticipating or Suffering the Loss of a Parent to Cancer: Helpful or Harmful?

Purpose: Assess mental health professionals’ attitudes regarding the timing and characteristics of therapeutic interventions for children whose parents have incurable cancer, and whether professionals would use artificial intelligence (AI) in these interventions. Methods: Professionals were surveyed about their therapeutic approaches to caring for children when parents have incurable cancer under different scenarios. Data from N = 294 (69% male, 72% white, 26% Latine, 56% rural or underserved communities) physicians, psychologists, social workers, hospital chaplains, community health workers, and others were analyzed. Attitudes surrounding the timing and characteristics of interventions across the parent’s cancer journey were compared, including how professionals believed interventions should attend to dimensions of the child or family, and if, how, and when AI technology could be introduced. Results: Across 10 dimensions of childhood, (1) the child’s premorbid exposure to traumatic events, (2) a surviving parent’s presence, and (3) the child’s age were important factors to consider when making mental health care decisions in this context. The professionals reported being more likely to introduce therapeutic resources as early as possible in the parent’s illness (i.e., upon diagnosis). Regarding the use of AI, 87% foresaw its role in supporting children’s mental health. While 93.2% agreed that a grieving child could be helped by interacting with an AI-generated likeness of the deceased parent, when AI’s use was contextualized in providing support for a child who lost a parent to cancer, only 49% believed AI was appropriate. The participants were conflicted over when AI could be first introduced, either upon a parent’s illness diagnosis (19.4%), during a parent’s treatment (19.0%), or as part of a parent’s hospice care (12.6%). None believed it to be appropriate following the loss of the parent to cancer. Conclusions: AI is increasingly present in children’s daily lives and quickly infiltrating health care with widely accessible mental health chatbots. Concerns about privacy, the accuracy of information, and the anthropomorphism of AI tools by children give professionals pause before introducing such technology. Proceeding with great caution is urged until more is known about the impact of AI on children’s mental health, grief, and psychological well-being in the context of parental cancer.

Sentiment Analysis‐Based Model for Monitoring User Engagement With Mental Health Chatbots

ABSTRACTMental health challenges, particularly among youth, are compounded by stigma and limited access to professional care. This has driven demand for scalable digital solutions like chatbots. This study introduces a sentiment analysis‐based model to assess user satisfaction with mental health chatbots, analyzing 82 102 reviews from six popular apps on Google Play and Apple's App Stores. A multi‐class sentiment classification of positive, negative, and neutral was implemented, enhanced by Synthetic Minority Over‐sampling Technique for class balancing, comparing five traditional machine learning models with Bidirectional Encoder Representations from Transformers, a transformer model. Random Forest achieved 98.18% accuracy among traditional models, while BERT outperformed all with 99.17% accuracy, surpassing prior benchmarks. Aspect‐based analysis revealed that Emotion and Usability drive positive feedback, while Reliability issues fuel negative sentiments, offering actionable insights for developers to enhance chatbot design. This work advances digital mental health research by integrating multi‐class classification, transformer models, and aspect‐based analysis, demonstrating a scalable framework for evaluating user feedback.

Is This Chatbot Safe and Evidence-Based? A Call for the Critical Evaluation of Generative AI Mental Health Chatbots

The proliferation of artificial intelligence (AI)–based mental health chatbots, such as those on platforms like OpenAI’s GPT Store and Character. AI, raises issues of safety, effectiveness, and ethical use; they also raise an opportunity for patients and consumers to ensure AI tools clearly communicate how they meet their needs. While many of these tools claim to offer therapeutic advice, their unregulated status and lack of systematic evaluation create risks for users, particularly vulnerable individuals. This viewpoint article highlights the urgent need for a standardized framework to assess and demonstrate the safety, ethics, and evidence basis of AI chatbots used in mental health contexts. Drawing on clinical expertise, research, co-design experience, and the World Health Organization’s guidance, the authors propose key evaluation criteria: adherence to ethical principles, evidence-based responses, conversational skills, safety protocols, and accessibility. Implementation challenges, including setting output criteria without one “right answer,” evaluating multiturn conversations, and involving experts for oversight at scale, are explored. The authors advocate for greater consumer engagement in chatbot evaluation to ensure that these tools address users’ needs effectively and responsibly, emphasizing the ethical obligation of developers to prioritize safety and a strong base in empirical evidence.

AI in mental healthcare: hope or hype?

Artificial Intelligence (AI) holds promise for addressing significant challenges in mental healthcare, such as workforce shortages, waiting lists, and the need for personalized diagnostics and treatments. However, current high expectations contrast with the complexities revealed by research on clinical implementations. This paper critically examines two clinical applications of AI in mental healthcare: clinical decision support and mental health chatbots. AI-driven decision support tools, despite their capacity to predict treatment outcomes, face ethical and practical challenges, including fairness concerns, limited outcome measurability, and patient perspectives often overlooked during development. Similarly, mental health chatbots, while enhancing accessibility and supporting therapeutic processes, raise concerns regarding patient autonomy, accountability, and safety. Increasing cases of AI chatbot misuseleading toharmful consequences highlight the need for robust ethical guidelines and rigorous validation before clinical implementation. AI's integration in psychiatry thus requires careful consideration of ethical implications, patient-centered design, and recognition of AI's limitations to ensure improvements genuinely enhance patient care.

SYSTEMATIC REVIEW ON AI IN EMOTIONAL INTELLIGENCE AND PSYCHOLOGICAL EDUCATION

Artificial Intelligence (AI) has emerged as a transformative force in emotional intelligence (EI) and psychological education, offering new pathways for emotion-aware learning, mental health support, and personalised interventions. This systematic review synthesises findings from 30 peer-reviewed studies published between 2015 and 2025 to evaluate the role of AI in assessing, enhancing, and supporting emotional and psychological competencies in educational settings. The review categorises AI applications into five thematic domains: emotion recognition technologies (e.g., EEG, facial analysis), adaptive learning systems, mental health chatbots, social-emotional learning (SEL) platforms, and ethical frameworks. Evidence indicates that AI tools can improve learners' emotional regulation, self-awareness, empathy, and psychological engagement. At the same time, critical challenges persist, including algorithmic bias, cross-cultural emotion misinterpretation, privacy concerns, and a lack of standardised evaluation metrics. The review further explores how these technologies align with and challenge existing theories such as Emotional Intelligence Theory, Affective Computing, Self-Regulated Learning, and Constructivism. A future research roadmap is proposed to guide interdisciplinary development, emphasising culturally inclusive design, longitudinal validation, hybrid human-AI emotional scaffolding, and ethical governance. This work contributes to a deeper understanding of AI’s evolving role as a co-regulator of emotional learning and a catalyst for mental well-being in education.

Validation of an Artificial Intelligence-Powered Virtual Assistant for Emergency Triage in Neurology.

Neurological emergencies pose significant challenges in medical care in resource-limited countries. Artificial intelligence (AI), particularly health chatbots, offers a promising solution. Rigorous validation is required to ensure safety and accuracy. Our objective is to evaluate the diagnostic safety and effectiveness of an AI-powered virtual assistant (VA) designed for the triage of neurological pathologies. The performance of an AI-powered VA for emergency neurological triage was tested. Ten patients over 18 years old with urgent neurological pathologies were selected. In the first stage, 9 neurologists assessed the safety of the VA using their clinical records. In the second stage, the assistant's accuracy when used by patients was evaluated. Finally, VA performance was compared with ChatGPT 3.5 and 4. In stage 1, neurologists agreed with the VA in 98.5% of the cases for syndromic diagnosis, and in all cases, the definitive diagnosis was among the top 5 differentials. In stage 2, neurologists agreed with all diagnostic parameters and recommendations suggested by the assistant to patients. The average use time was 5.5 minutes (average of 16.5 questions). VA showed superiority over both versions of ChatGPT in all evaluated diagnostic and safety aspects ( P <0.0001). In 57.8% of the evaluations, neurologists rated the VA as "excellent" (suggesting adequate utility). In this study, the VA showcased promising diagnostic accuracy and user satisfaction, bolstering confidence in further development. These outcomes encourage proceeding to a comprehensive phase 1/2 trial with 100 patients to thoroughly assess its "real-time" application in emergency neurological triage.

User Needs and Benefits of Mental Health Chatbots: Text-Mining Analysis of Mobile Apps Reviews

This study investigates user needs and perceived benefits of mental health chatbots by analyzing user reviews from the AppStore and Google Play Store for three different chatbots. Text-mining analysis reveals a strong preference for social interactivity, with users frequently mentioning terms like “talking” and “human,” which suggests that users value a high degree of social and hedonic expectations from these chatbots. There is also a noticeable enthusiasm for innovation and technology, as indicated by mentions of terms such as “recent” and “newest.” Cluster analysis results further highlight the importance of psychoeducation, which generally evokes positive sentiments, while perceptions around pricing show a wider range of responses. Overall, our findings underscore the importance of social features in mental health chatbots and offer valuable insights on pricing preference. These insights are intended to support developers and mental health professionals in enhancing the effectiveness and appeal of these applications.

AI-human interactions in healthcare: exploring users’ post-adoption behaviors of AI mental health chatbots

ABSTRACT The rise in mental health challenges has spurred innovative technological solutions, including AI-based chatbots for mental care management. This study explores how individuals engage with these chatbots for health decision-making, examining relationships between utilitarian and hedonic values, satisfaction with well-being outcomes, IT identity, social norms, and post-adoption behaviours. Using IT identity theory and a quantitative approach, we surveyed 309 current users of AI-based chatbots in the US. Structural equation modeling (SEM) was employed to analyse the data. Results show that both utilitarian and hedonic values positively influence user satisfaction with well-being outcomes. Satisfaction, in turn, shapes IT identity, reflecting personalised engagement with technology. IT identity positively influences behavioural intentions, including exploring new features, word of mouth, and continued use of AI chatbots. The study reveals mixed moderating effects of social norms, with significant moderation for word of mouth but not for behavioural intentions. Mediation effects of satisfaction in the relationships between values and IT identity provide insights into how technology becomes integral to individual identities and well-being routines. This research contributes to both theoretical understanding and practical applications of AI chatbots in mental health and well-being. It offers a robust framework for future research and interventions in AI-enabled healthcare.

Chatbot -assisted self-assessment (CASA): Co-designing an AI -powered behaviour change intervention for ethnic minorities.

The digitalisation of healthcare has provided new ways to address disparities in sexual health outcomes that particularly affect ethnic and sexual minorities. Conversational artificial intelligence (AI) chatbots can provide personalised health education and refer users for appropriate medical consultations. We aimed to explore design principles of a chatbot-assisted culturally sensitive self-assessment intervention based on the disclosure of health-related information. In 2022, an online survey was conducted among an ethnically diverse UK sample (N = 1,287) to identify the level and type of health-related information disclosure to sexual health chatbots, and reactions to chatbots' risk appraisal. Follow-up interviews (N = 41) further explored perceptions of chatbot-led health assessment to identify aspects related to acceptability and utilisation. Datasets were analysed using one-way ANOVAs, linear regression, and thematic analysis. Participants had neutral-to-positive attitudes towards chatbots and were comfortable disclosing demographic and sensitive health information. Chatbot awareness, previous experience and positive attitudes towards chatbots predicted information disclosure. Qualitatively, four main themes were identified: "Chatbot as an artificial health advisor", "Disclosing information to a chatbot", "Ways to facilitate trust and disclosure", and "Acting on self-assessment". Chatbots were acceptable for health self-assessment among this sample of ethnically diverse individuals. Most users reported being comfortable disclosing sensitive and personal information, but user anonymity is key to engagement with chatbots. As this technology becomes more advanced and widely available, chatbots could potentially become supplementary tools for health education and screening eligibility assessment. Future research is needed to establish their impact on screening uptake and access to health services among minoritised communities.

Improving Workplace Well-being in Modern Organizations: A Review of Large Language Model-based Mental Health Chatbots

The global rise in mental disorders, particularly in workplaces, necessitated innovative and scalable solutions for delivering therapy. Large Language Model (LLM)-based mental health chatbots have rapidly emerged as a promising tool for overcoming the time, cost, and accessibility constraints often associated with traditional mental health therapy. However, LLM-based mental health chatbots are in their nascency, with significant opportunities to enhance their capabilities to operate within organizational contexts. To this end, this research seeks to examine the role and development of LLMs in mental health chatbots over the past half-decade. Through our review, we identified over 50 mental health-related chatbots, including 22 LLM-based models targeting general mental health, depression, anxiety, stress, and suicide ideation. These chatbots are primarily used for emotional support and guidance but often lack capabilities specifically designed for workplace mental health, where such issues are increasingly prevalent. The review covers their development, applications, evaluation, ethical concerns, integration with traditional services, LLM-as-a-Service, and various other business implications in organizational settings. We provide a research illustration of how LLM-based approaches could overcome the identified limitations and also offer a system that could help facilitate systematic evaluation of LLM-based mental health chatbots. We offer suggestions for future research tailored to workplace mental health needs.

Unleashing the potential of chatbots in mental health: bibliometric analysis.

The proliferation of chatbots in the digital mental health sector is gaining momentum, offering a promising solution to address the pressing shortage of mental health professionals. By providing accessible and convenient mental health services and support, chatbots are poised to become a primary technological intervention in bridging the gap between mental health needs and available resources. This study undertakes a thorough bibliometric analysis and discourse on the applications of chatbots in mental health, with the objective of elucidating the underlying scientific patterns that emerge at the intersection of chatbot technology and mental health care on a global scale. The bibliometric software Biblioshiny and VOSviewer were used to conduct a comprehensive analysis of 261 scientific articles published in the Web of Science Core Collection between 2015 and 2024. Publications distribution are analyzed to measure productivity on countries, institutions, and sources. Scientific collaboration networks are generated to analyze the influence as well as communications between countries and institutions. Research topics and trends are formulated by using a keyword co-occurrence network. Over the last decade, researches on utilization of chatbots in mental health has appeared to be increasing steadily at an annual rate of 46.19%. The United States have made significant contributions to the development and expansion of publications, accounting for 27.97% of the total research output with 2452 citation counts. England came second to the US in terms of publications and citations, and followed by Australia, China, and France. National Center for Scientific Research in France ranked first among all institutions, followed by Imperial College London and University of Zurich. The number of articles published in Journal of Medical Internet Research was exceptionally high, accounting for 12.26% of the total number of articles, and JMIR Mental Health is the most influential publication sources in terms of average citations per article. Collaboration among universities in the USA, United Kingdom, Switzerland, and Singapore demonstrated a high level. The keyword co-occurrence network highlights the prominent techniques in this multidisciplinary area and reveals 5 research topics, showing a significant overlap between clusters. High-frequency terms such as "ChatGPT", "machine learning", and "large language models" underscore the current state of research, highlighting the cutting-edge advancements and frontiers in this field. This study provides an in-depth analysis of the most prominent countries, institutions, publications, collaboration status, and research topics associated with utilization of chatbots in mental health over the last decade. It offers insights to mental health professionals without an AI background and individuals interested in the development of mental health chatbots. The findings suggest that chatbots hold a significant role in promoting mental health well-being and exhibit considerable potential in demonstrating empathy, curiosity, understanding, and collaborative capabilities with users.

Your artificial intelligence will see you now: Why nurse practitioners remain irreplaceable

ABSTRACT Artificial intelligence (AI) has transformed health care. Artificial intelligence technologies, such as advanced imaging algorithms, diagnostic tools, and mental health chatbots, have revolutionized patient care by enhancing diagnostic accuracy, personalizing treatment plans, and streamlining administrative tasks. However, despite these advancements, AI falls short in areas where nurse practitioners (NPs) excel. Nurse practitioners possess essential human attributes such as empathy, nuanced understanding, and ethical reasoning that AI cannot currently replicate. They excel at recognizing subtle mood changes, understanding social determinants of health, and navigating complex ethical dilemmas. I argue that although AI can support and enhance health care delivery, it cannot replace the indispensable human touch provided by NPs. The irreplaceable role of NPs in offering holistic, compassionate care underscores the need for a balanced integration of AI, to ensure it complements rather than replaces the human elements crucial to effective patient care.

Exploring Client Perceptions on Gaining Infant Feeding Information Through the Texas Women, Infants, and Children (WIC) Chatbot.

The modernization of the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) program is a priority. The Texas WIC chatbot, Maya, streamlines client interactions through dialog-based responses. This qualitative study explored client capabilities, motivations, and opportunities for seeking nutrition information about breastfeeding, formula feeding, infant feeding safety, adequacy of infant feeding, and complementary feeding via a chatbot. A team conducted in-depth semi-structured interviews with Texas WIC clients (n = 19 women). All interviews were transcribed and subjected to a two-coder, four-phase process utilizing a theory-based codebook. Codes were compiled and thematically categorized. Identified themes included (1) motivations through necessity or resource availability, (2) client capabilities and Maya usability, and (3) opportunities for connection, support, and encouragement. Texas WIC clients that participated in this study expressed motivations, capabilities, and opportunities to engage with nutrition information through Maya. They described Maya as a favorable resource for behavior changes, and a trusted source of nutrition information, citing the credibility of WIC and reliability of the chatbot. The findings may inform future research and development of public health chatbots. Additional research is required to explore how different factors such as language and technology usage may impact client capabilities, motivations, and opportunities to seek nutrition information with regard to infant feeding.

How does a mental health chatbot work? A ‘conversation design’ concept of mental health intervention

Mental health chatbots, along with computerised treatment in general, have gradually entered the realm of acceptability. This article looks at a chatbot called ReMind. It begins with an overview of the development of computerised mental health interventions, drawing links between the invention of cognitive behavioural therapy and automated therapy. The focus then moves to analysis of ReMind's conversational system. The bot acts as a sympathetic guide which directs the user to mental health activities, and as we will see, a concept of mental health is produced through this system which is unique to the way that chatbot conversations are designed. The bot is tasked with interpreting the user's typed inputs in terms of keywords. Keywords are identified in terms of predetermined tags (‘sad’, ‘can’t sleep’, ‘relationship’, etc.). The bot then ‘predicts’ what the most appropriate response should be. ‘Prediction’ will be focused on to draw out how users interact with the bot in terms of conforming to the conversational logic imposed by the bot. This article illustrates a concept of mental health which is produced in the conversational dynamic between the users and the bot, and in how these conversations are directed on an abstract level by the ReMind company. The article concludes that the tautological cycle that the ReMind app facilitates represents a model of mental health which is correspondingly tautological in terms of users adopting an attitude of adaptation. This article addresses an urgent need to critically assess new forms of mental health interventions, and especially, technologically informed interventions.

Recommendations for Mental Health Chatbot Conversations: An Integrative Review.

To identify and synthesise recommendations and guidelines for mental health chatbot conversational design. Integrative review. Suitable publications presenting recommendations or guidelines for mental health conversational design were included. The quality of included publications was assessed using Joanna Briggs Institute Critical Appraisal Tools. Thematic analysis was conducted. Primary searches limited to last 10 years were conducted in PubMed, Scopus, ACM Digital Library and EBSCO databases including APA PsycINFO, CINAHL, APA PsycArticles and MEDLINE in February 2023 and updated in October 2023. A secondary search was conducted in Google Scholar in May 2023. Of 1684 articles screened, 16 publications were selected. Three overarching themes were developed: (1) explicit knowledge about chatbot design and domain, (2) knowing your audience and (3) creating a safe space to engage. Results highlight that creating pleasant and effective conversations with a mental health chatbot requires careful and professional planning in advance, defining the target group and working together with it to address its needs and preferences. It is essential to emphasise the pleasant user experience and safety from both technical and psychological perspectives. Recommendations for mental health chatbot conversational design have evolved and become more specific in recent years. Recommendations set high standards for mental health chatbots. To meet that, co-design, explicit knowledge of the user needs, domain and conversational design is needed. Mental health professionals participating in chatbot development can utilise this review. The results can also inform technical development teams not involving healthcare professionals directly. Knowledge of developing mental health chatbot conversations appears scattered. In mental health chatbots, features that enhance the chatbot's ability to meet users' needs and increase safety should be considered. This review is useful for developers of mental health chatbots and other health applications used independently. This integrative review was reported according to PRISMA guidelines, as applicable. No patient or public contribution.

Development and Evaluation of a Mental Health Chatbot Using ChatGPT 4.0: Mixed Methods User Experience Study With Korean Users.

Mental health chatbots have emerged as a promising tool for providing accessible and convenient support to individuals in need. Building on our previous research on digital interventions for loneliness and depression among Korean college students, this study addresses the limitations identified and explores more advanced artificial intelligence-driven solutions. This study aimed to develop and evaluate the performance of HoMemeTown Dr. CareSam, an advanced cross-lingual chatbot using ChatGPT 4.0 (OpenAI) to provide seamless support in both English and Korean contexts. The chatbot was designed to address the need for more personalized and culturally sensitive mental health support identified in our previous work while providing an accessible and user-friendly interface for Korean young adults. We conducted a mixed methods pilot study with 20 Korean young adults aged 18 to 27 (mean 23.3, SD 1.96) years. The HoMemeTown Dr CareSam chatbot was developed using the GPT application programming interface, incorporating features such as a gratitude journal and risk detection. User satisfaction and chatbot performance were evaluated using quantitative surveys and qualitative feedback, with triangulation used to ensure the validity and robustness of findings through cross-verification of data sources. Comparative analyses were conducted with other large language models chatbots and existing digital therapy tools (Woebot [Woebot Health Inc] and Happify [Twill Inc]). Users generally expressed positive views towards the chatbot, with positivity and support receiving the highest score on a 10-point scale (mean 9.0, SD 1.2), followed by empathy (mean 8.7, SD 1.6) and active listening (mean 8.0, SD 1.8). However, areas for improvement were noted in professionalism (mean 7.0, SD 2.0), complexity of content (mean 7.4, SD 2.0), and personalization (mean 7.4, SD 2.4). The chatbot demonstrated statistically significant performance differences compared with other large language models chatbots (F=3.27; P=.047), with more pronounced differences compared with Woebot and Happify (F=12.94; P<.001). Qualitative feedback highlighted the chatbot's strengths in providing empathetic responses and a user-friendly interface, while areas for improvement included response speed and the naturalness of Korean language responses. The HoMemeTown Dr CareSam chatbot shows potential as a cross-lingual mental health support tool, achieving high user satisfaction and demonstrating comparative advantages over existing digital interventions. However, the study's limited sample size and short-term nature necessitate further research. Future studies should include larger-scale clinical trials, enhanced risk detection features, and integration with existing health care systems to fully realize its potential in supporting mental well-being across different linguistic and cultural contexts.

How does the degree of anthropomorphism of health chatbots affect the public's willingness to seek help from them? Empirical research using HBM

Currently, health chatbots have a wide range of applications in the medical field. However, most of the related research centers around self-diagnosis at the user's physical level, while the psychological level is rarely discussed. Unlike real doctors, health chatbots are more accessible, more convenient and less stigmatizing. Consequently, this study develops a structural equation model on the degree of anthropomorphism of chatbots and willingness to ask for help from them based on the health belief model, the technology acceptance model, and the privacy computing theory. It contributes to complementing the lack of research on human-robot communication in health communication. The results show that an increase in the degree of chatbot anthropomorphism significantly increases users' perceived benefits and reduces privacy concerns, thus increasing their willingness to seek help from it. Interestingly, increased anthropomorphism enhances users' perceived severity and perceived susceptibility, which may be due to the specificity of mental health problems, where the more the chatbot resembles a real person, the more likely it is to aggravate users' tension and anxiety. This study also found that Chinese users generally have a low level of acceptance of psychological counseling chatbots, and chatbots are not their first choice for help when experiencing psychological problems. In the future, the professionalism of the psychological counseling function of health chatbots can be further improved, and at the same time well publicized for the benefit of more people troubled by psychological problems.