Sage Journals: Discover world-class research

Abstract

Background

The integration of artificial intelligence (AI) into therapeutic practices has introduced new possibilities for enhancing mental health interventions. Among these innovations, AI-driven art therapy has emerged as a promising approach, combining the expressive and healing aspects of traditional art therapy with the adaptive and analytical strengths of AI. This modality shows particular potential in supporting individuals with autism spectrum disorder (ASD), mental health challenges, and those seeking emotional well-being.

Aim

This umbrella review aimed to synthesize and evaluate existing evidence on the application, effectiveness, and implementation of AI-driven art therapy in improving outcomes for individuals with ASD, mental health conditions, and emotional distress.

Methods

A comprehensive search was conducted across multiple databases—including PubMed, Scopus, Web of Science, PsycINFO, IEEE Xplore, and ACM Digital Library—to identify systematic and scoping reviews that adhered to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The methodological quality of included reviews was assessed using the Joanna Briggs Institute (JBI) checklist. Thematic analysis was applied to extract and synthesize recurring patterns and core themes across studies.

Results

The findings revealed significant therapeutic benefits of AI-driven art therapy, including enhanced communication and emotional regulation in individuals with ASD, and reduced symptoms of anxiety, depression, and Post-Traumatic Stress Disorder (PTSD) in mental health populations. Additionally, the platforms improved emotional well-being through personalized engagement, self-reflection, and increased autonomy. However, technological barriers, ethical concerns, and accessibility issues were noted as key limitations.

Conclusions

AI-driven art therapy represents a transformative and accessible tool in modern therapeutic practices. While challenges remain, its potential to support diverse populations through individually tailored and engaging interventions makes it a valuable complement to traditional mental health care.

Keywords

AI-driven art therapy autism spectrum disorder (ASD)mental health interventions emotional well-being artificial intelligence in therapy

Introduction

The integration of AI into therapeutic contexts has opened up innovative possibilities, especially in fields that benefit from personalization and emotional insight.¹ Among these, AI-driven art therapy is emerging as a particularly promising approach.² This method combines the creative, healing aspects of traditional art therapy with the analytical and adaptive capabilities of AI,^3,4 creating a responsive and interactive space for emotional expression and psychological support.⁵ In this review, AI-driven art therapy refers to the use of AI tools such as generative art systems, interactive digital platforms, emotional analysis applications, and recommendation algorithms that personalize and support art-based therapeutic practices. It holds significant potential for populations such as individuals with ASD,⁶ those experiencing mental health issues like anxiety or depression, and people simply looking to enhance their emotional well-being in a more intuitive and creative way.^7,8

Art therapy has long served as a channel for emotional expression, allowing individuals to communicate complex or unspoken feelings through visual and creative means such as painting, sketching, or digital art.^9,10 The introduction of AI enhances this by offering real-time support and personalization.¹¹ AI tools can interpret patterns in a person's behavior or artistic choices, suggest tailored interventions, and provide adaptive prompts that respond to the user's emotional state.^12,13 This makes the therapy more inclusive and accessible, even for those who may be uncomfortable with or excluded from traditional therapeutic environments.^14–16 Particularly for individuals with ASD, who often face difficulties with verbal communication and sensory regulation,¹⁷ these AI-supported platforms create structured, low-pressure spaces that are both engaging and easy to navigate.¹⁸

Mental health is another area where AI-based art therapy is showing considerable promise.¹⁹ Globally, depression affects about 280 million people, anxiety disorders about 300 million,²⁰ and PTSD around 4% of the population.²¹ AI-enhanced platforms can offer safe environments where individuals are free to express emotions,²² reflect on their inner experiences, and build coping strategies without judgment.^23,24 Features such as mood tracking, virtual creative environments, and intelligent feedback loops allow for ongoing, self-paced therapeutic engagement.²⁵ Moreover, these platforms empower users by offering a sense of control and self-awareness—essential elements in both recovery and emotional growth.^26–28 They not only facilitate expression but also promote emotional resilience and empowerment, making them relevant even for those outsides of formal mental health diagnoses.^29,30

Despite these advancements, several questions remain about the reach and effectiveness of AI-driven art therapy.³¹ The field is still developing, and existing research is often fragmented, with studies focusing on specific disorders, age groups, or types of AI technologies.^32,33 Some employ simple algorithms, while others explore advanced machine learning (ML) systems within immersive or traditional art-based environments.^34,35 As the landscape continues to evolve, it is crucial to consolidate what is already known. The aim of this review is to synthesize and evaluate existing evidence from reviews on the application, effectiveness, and implementation of AI-driven art therapy in supporting individuals with autism, mental health conditions, and emotional well-being.

Methodology

Research design

The current study employed an umbrella review methodology, strategically chosen for its capability to integrate and synthesize evidence from various systematic, scoping, and PRISMA-compliant review papers comprehensively. This approach facilitated the identification and exploration of critical themes and existing knowledge gaps regarding the effectiveness of AI-driven art therapy in enhancing ASD, mental health, and emotional well-being. The umbrella review framework provided a structured methodological foundation to consolidate diverse findings systematically, offering a comprehensive overview of the existing evidence landscape.

Literature search strategy

An extensive systematic search of the literature was conducted across multiple prominent scholarly databases, including PubMed, Scopus, Web of Science, PsycINFO, IEEE Xplore, and ACM Digital Library. Search strategies were developed using targeted combinations of keywords, specifically: “artificial intelligence,” “AI-driven art therapy,” “autism spectrum disorder,” “mental health,” “emotional well-being,” and related terms. Boolean operators were applied in detail to improve both precision and comprehensiveness. The operator “AND” was used to combine different concepts, ensuring that retrieved articles included all specified terms (e.g., “artificial intelligence AND autism spectrum disorder”). The operator “OR” was used to capture synonyms and related expressions, broadening the scope of retrieval (e.g., “AI-driven art therapy OR digital art therapy”). The operator “NOT” was used to exclude irrelevant material that might otherwise appear in search results (e.g., “artificial intelligence NOT robotics”). Each database's search strategy was tailored to its indexing conventions and retrieval systems, maximizing the breadth and accuracy of relevant review identification.

Inclusion and exclusion criteria

Explicit and comprehensive inclusion criteria were established to guide the selection of reviews. Eligible studies included peer-reviewed systematic reviews, scoping reviews, integrative reviews, systematic reviews with meta-analysis or meta-synthesis, and reviews explicitly adhering to PRISMA guidelines. Only articles written in English and directly addressing AI-driven art therapy interventions targeting populations with autism, mental health disorders, or emotional well-being challenges were considered. To ensure methodological rigor and clarity, reviews were excluded if they were non-peer-reviewed, grey literature, abstracts, conference proceedings, editorials, commentaries, opinion pieces, or narrative reviews lacking clearly defined methodological frameworks. This approach ensured the inclusion of only well-structured and methodologically robust review articles capable of contributing reliable and relevant insights to the umbrella review.

Screening and selection procedure

The article selection process followed a structured, multiphase approach comprising initial title and abstract screening followed by comprehensive full-text assessment. Two independent reviewers systematically evaluated titles and abstracts against the predefined inclusion criteria to exclude irrelevant studies initially. Subsequently, full-text versions of the remaining articles underwent thorough scrutiny by both reviewers to verify eligibility conclusively. Disagreements or discrepancies identified at any screening phase were systematically documented and resolved through detailed discussions or consultation with a third reviewer, ensuring objective and reliable decision-making.

Quality assessment

To ensure methodological rigor, the quality of the included systematic reviews and scoping reviews was appraised using the Joanna Briggs Institute (JBI) critical appraisal checklist for systematic reviews and research syntheses³⁶ (Table 1). The JBI tool evaluates key aspects such as the clarity and relevance of the review question, the appropriateness of the inclusion criteria, the thoroughness of the search strategy, the transparency of study selection and data extraction methods, and the reliability of data synthesis and interpretation. Two independent reviewers conducted the critical appraisal of each review article. Any differences in their evaluations were addressed through collaborative discussion, and when an agreement could not be reached, input from a third reviewer was sought to finalize the assessment. This process ensured a consistent, transparent, and rigorous assessment of the methodological quality of the included literature. Although the JBI 10-item tool was applied to categorize studies into high, moderate, or low methodological quality, the appraisal revealed that all included reviews were rated as high quality, thereby reinforcing the robustness and reliability of the synthesized evidence.

Table 1.

JBI quality assessment.

References	Authors and year	Clearly defined review question	Appropriate inclusion criteria	Comprehensive search strategy	Transparent selection methods	Transparent data extraction	Critical appraisal conducted	Appropriate synthesis methods	Conclusions supported by data	Publication bias assessed	Recommendations relevant to practice
³⁷	Banos et al. (2024)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
³⁸	Rebelo et al. (2023)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
³⁹	Thenral and Annamalai (2020)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴⁰	Bouchouras and Kotis (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴¹	Cruz-Gonzalez et al. (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴²	Tay et al. (2024)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴³	de Belen et al. (2020)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴⁴	Higgins et al. (2023)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴⁵	Milasan and Scott-Purdy (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴⁶	Alqaysi et al. (2022)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴⁷	Scoglio et al. (2019)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴⁸	Le Glaz et al. (2021)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁴⁹	Abd-alrazaq et al. (2023)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵⁰	Auf et al. (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵¹	Milne-Ives et al. (2022)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵²	Cavus et al. (2021)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵³	Moon et al. (2019)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵⁴	Rogan et al. (2024)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵⁵	Rahsepar Meadi et al. (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵⁶	Scherbakov et al. (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵⁷	Cecula et al. (2021)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵⁸	Bhatt et al. (2022)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁵⁹	Ding et al. (2024)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁶⁰	Su et al. (2020)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁶¹	Islam et al. (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁶²	Alabdulkareem et al. (2022)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁶³	Solek et al. (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁶⁴	Kohli et al. (2022)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁶⁵	Kolding et al. (2025)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
⁶⁶	Li et al. (2023)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Data extraction process

Data extraction was conducted systematically using a customized, structured template specifically designed to capture essential details relevant to the objectives of the umbrella review. Extracted data points included authors, publication year, objectives, study design, types of AI-driven art therapy interventions discussed, targeted populations, evaluated outcomes, and key findings (Table 2). To maintain accuracy and minimize bias, two independent reviewers carried out the data extraction process and resolved any discrepancies through collaborative discussion, ensuring both thoroughness and reliability.

Table 2.

Summary of included reviews.

References	Authors and year	Study design	Objectives	AI-driven interventions discussed	Targeted populations	Evaluated outcomes	Key findings
³⁷	Banos et al. (2024)	Systematic review	To systematically review sensing technologies and machine learning methods for human emotion recognition (HER) in individuals with autism, identifying used techniques, challenges, and trends.	Facial expression recognition, speech analysis, body movement tracking, EEG/EMG sensing, multimodal approaches, emotion-learning apps/games, robotic platforms, augmented/virtual reality tools.	Individuals with autism spectrum disorder (ASD), including children, adolescents, and adults; also included comparisons with neurotypical individuals and subtypes such as high-functioning, mild, and Asperger's.	Model performance (accuracy, sensitivity), emotional recognition improvement, communication enhancement, engagement levels, and ethical adherence (e.g., data privacy).	Facial cues dominate HER; supervised ML (SVM) most common; deep learning underused; poor demographic/ethical reporting; lack of autism-specific datasets; need for context-aware and personalized systems.
³⁸	Rebelo et al. (2023)	Scoping review	To characterize the impact of AI on the tasks of mental healthcare workers, describe how these systems impact the tasks, and identify which tasks are affected.	AI chatbots, machine learning algorithms, deep learning systems (e.g., CNNs), symbolic AI, fuzzy logic systems for assessment, therapy support, patient monitoring, emotion detection, and diagnosis.	Mental healthcare professionals (psychologists, psychiatrists, therapists/counselors); Indirectly includes patients with mental health conditions like depression, schizophrenia, autism, etc.	Task performance (assessment, therapy, monitoring, documentation, prescription), emotional state detection, treatment prediction, therapist-patient interaction analysis, professional perspectives on AI integration.	AI systems primarily assist rather than replace mental healthcare workers. Most effective in assessment and monitoring tasks. Ethical concerns and dataset limitations noted. Professionals need training to adapt to hybrid intelligence settings.
³⁹	Thenral and Annamalai (2020)	Scoping review	To explore the potential of telepsychiatry and AI in mental health in post-COVID-19 India and discuss their integration and applications in psychiatric care.	AI-enabled telepsychiatry, asynchronous and synchronous models, web-based interactions, mobile applications, virtual platforms, knowledge-based systems, emotion analysis, acoustic and biosignal assessment.	Mental health patients in India during and after COVID-19; indirectly mental health professionals and policymakers.	Adoption feasibility, technological features, model effectiveness, clinician readiness, changes in health-seeking behavior, and system-level opportunities and constraints.	Telepsychiatry and AI can help bridge the mental health service gap, especially in India, but require indigenous development, validation, and collaboration between psychiatrists and technologists.
⁴⁰	Bouchouras and Kotis (2025)	Systematic review	To review methodologies involving AI, IoT, and sensor-based technologies for ASD detection and examine their potential to enhance diagnostic accuracy and personalization.	Wearable sensors, eye-tracking systems, virtual reality, neuroimaging, EEG devices, smart environments, serious games, vocal analysis, facial and head movement recognition, IoT-based monitoring systems, microbiome analysis.	Children with autism spectrum disorder (ASD), including toddlers and school-aged children; clinicians and researchers in neurodevelopmental diagnostics.	Diagnostic accuracy, sensitivity/specificity of algorithms, real-time monitoring capabilities, behavioral and neurological insights, technological integration, and scalability.	AI, IoT, and sensor-based technologies show promise for non-invasive, real-time, and personalized ASD detection, but need further validation and integration with clinical standards.
⁴¹	Cruz-Gonzalez et al. (2025)	Systematic review	To analyze the role of AI in diagnosis, monitoring, and intervention within mental health care, including limitations and ethical concerns.	Support vector machine, random forest, AI chatbots, deep learning, AI-assisted diagnosis, monitoring tools, and interventions.	Individuals with mental health disorders including depression, schizophrenia, anxiety, OCD, and suicide risk.	AI model accuracy in diagnosis, symptom monitoring, treatment prediction, ethical reporting, and model performance.	AI tools are promising in mental health care, especially in diagnosis and intervention, though ethical gaps and data limitations persist.
⁴²	Tay et al. (2024)	Scoping review	To assess the accuracy of AI models in predicting clinical outcomes in psychotic disorders and identify relevant predictors.	Machine learning models including SVM, RF, logistic regression, and linear regression applied to clinical, biological, and imaging data.	Patients with psychotic disorders (schizophrenia, schizoaffective disorder, delusional disorders).	Prediction accuracy (AUC 0.58–0.95), relevant predictors (demographics, clinical course, neuroimaging, treatment	AI models show good predictive performance but lack consistency across methods; integration of complex predictors is necessary.
⁴³	de Belen et al. (2020)	Systematic review (2009–2019 studies)	To examine how computer vision analysis has been useful in autism spectrum disorder (ASD) research, including diagnosis, therapy, and broader applications; categorize studies by biomarkers; review ASD datasets; outline research directions.	Computer vision methods: MRI/fMRI analysis, facial expression and emotion recognition (CNNs, SVM, HOG), eye gaze tracking, motor control/movement pattern analysis (CNN + LSTM, R-CNN), stereotyped behaviour detection, multimodal data integration.	Children and adults with ASD; control groups including typically developing (TD) individuals and sometimes developmental delay/intellectual disability populations.	Identification and quantification of ASD-related biomarkers; classification accuracy in ASD diagnosis; improvements in emotion recognition, behavioural monitoring, engagement; effectiveness of assistive/therapeutic technologies.	Computer vision effectively quantifies ASD biomarkers; provides objective, low-cost, and non-invasive diagnostic/therapeutic tools; MRI-based models achieved >90% accuracy; assistive technologies improved engagement; multimodal approaches enhanced diagnostic and therapeutic accuracy; reduces bias and increases reliability in autism research.
⁴⁴	Higgins et al. (2023)	Integrative Review	To evaluate the implementation and effectiveness of AI/ML-based decision support systems (DSS) in mental health care and identify barriers to adoption.	AI/ML-powered decision support systems used for prediction, diagnosis, and care optimization in clinical mental health settings.	Mental health clinicians and patients, particularly in systems using AI/ML for major depression and predictive support.	Barriers and enablers to clinician trust, user confidence, system transparency, and decision-making enhancement in DSS.	Trust and confidence are critical for DSS success; clinicians must be involved in system development to ensure real-world integration.
⁴⁵	Milasan and Scott-Purdy (2025)	Integrative review	To identify and synthesize evidence of AI-based applications relevant to enhancing mental health nursing practice.	Machine learning, natural language processing, deep learning, computer vision, digital phenotyping, virtual agents, predictive analytics.	Mental health nurses and patients in psychiatric hospitals, community mental health, and digital platforms.	AI impact on assessment, diagnosis, prediction, optimization, patient engagement, and perception in mental health nursing.	AI shows promise in enhancing nursing care through personalized assessment and prediction, but faces challenges in ethics and integration.
⁴⁶	Alqaysi et al. (2022)	Systematic review	To review literature on hybrid diagnosis of ASD using AI and ML based on multimedical tests and sociodemographic features.	AI/ML techniques including SVM, RF, KNN, CNN, RNN; integrating medical/lab tests with sociodemographic characteristics.	Children with autism spectrum disorder; focus on early diagnosis using diverse features (clinical, social, family).	Effectiveness of hybrid diagnostic models, feature selection, accuracy of prediction algorithms, and diagnosis challenges.	Hybrid AI approaches improve ASD diagnosis but lack integration of all relevant features; research should focus on feature synergy.
⁴⁷	Scoglio et al. (2019)	Systematic review	To synthesize empirical evidence on the use of socially assistive robots (SARs) in mental health and psychological well-being interventions.	Socially assistive robots (SARs) including Paro, NAO, CRECA, Betty, and Haptic Creature for companionship, stress reduction, motivation.	Adults and elderly patients in clinical and community settings; populations facing anxiety, depression, dementia, and stress.	User engagement, emotional response, motivation, mood, usability, stress reduction, and mental health improvement.	SARs show potential for improving mental health outcomes, but current evidence is limited in scope and lacks methodological rigor.
⁴⁸	Le Glaz et al. (2021)	Systematic review	To summarize and characterize studies that used machine learning and NLP techniques in mental health from methodological and technical perspectives.	Machine learning models, natural language processing tools, EHR-based classifiers, text mining, and social media analysis.	Patients with psychiatric disorders (e.g., depression, ASD, bipolar), emergency visitors, social media users.	Classification accuracy, symptom extraction, therapy evaluation, nosology refinement, NLP tool efficiency.	ML/NLP methods validate existing clinical hypotheses more than discovering new patterns; ethical concerns remain underexplored.
⁴⁹	Abd-alrazaq et al. (2023)	Scoping review	To explore the features of wearable AI used for anxiety and depression and identify its application areas and open research issues.	Wearable AI (e.g., Actiwatch, Fitbit, Empatica) collecting data like sleep, activity, heart rate; algorithms used included RF, SVM.	Adults aged 18–65 with anxiety and/or depression; some inclusion of healthy controls.	Diagnostic capabilities, real-time monitoring potential, performance of AI algorithms, gaps in treatment applications.	Wearable AI is promising for diagnosis and monitoring but not yet used for treatment; further evaluation needed.
⁵⁰	Auf et al. (2025)	Scoping review	To explore the use, implementation, and evaluation of AI systems in mental health decision-making support across various care contexts.	Diagnostic AI, treatment selection AI, and self-help AI; tools integrated in healthcare, self-care, and hybrid care settings.	Patients with mental health problems (depression, autism, suicide risk, substance use), healthcare providers, self-care users.	Decision-making support effectiveness, usability, integration challenges, trust, explainability, and human-AI interaction.	AI supports mental health decision-making but faces usability, integration, and evaluation challenges; empirical validation is limited.
⁵¹	Milne-Ives et al. (2022)	Scoping review	To review and map the current literature on AI/ML use in mobile apps for mental health support and identify strengths and gaps.	Mobile mental health apps using AI/ML for prediction, classification, personalization; includes chatbots and risk assessment.	Adults with or at risk of mental health conditions using AI-enabled mobile health apps.	Feasibility, usage trends, methodological gaps, and evidence on the effectiveness of AI-enabled mental health apps.	AI in mental health apps is feasible but under-researched; studies are early-phase with limited rigor and small samples.
⁵²	Cavus et al. (2021)	Systematic review	To systematically review ML models for behavioral ASD assessment and evaluate their readiness for real-life clinical implementation.	ML models using behavioral features (e.g., AQ-10, ADOS, SRS); feature selection and classification algorithms like RF, SVM, ANN.	Children with ASD; focus on early diagnosis using behavioral data and machine learning tools.	Classification accuracy, sensitivity, specificity, feature importance, real-world readiness of ML systems.	ML-based behavioral models show strong performance but face data challenges and conceptual alignment issues for clinical use.
⁵³	Moon et al. (2019)	Systematic review and meta-analysis	To estimate diagnostic accuracy of ML algorithms for ASD using brain MRI, through meta-analysis of sensitivity and specificity.	Machine learning algorithms (e.g., deep neural networks, support vector machines) were applied to neuroimaging data (sMRI, fMRI) for ASD diagnosis, with features extracted from brain imaging datasets	Individuals with ASD (preschoolers to adults) and neurotypical controls; data from neuroimaging studies.	Pooled sensitivity/specificity (sMRI: 0.83/0.84), subgroup heterogeneity, performance based on algorithm and imaging type.	ML algorithms show good accuracy for ASD diagnosis (especially with sMRI); however, validation and standardization are needed.
⁵⁴	Rogan et al. (2024)	Systematic review with meta-synthesis	To synthesize mental health care professionals’ views on passive sensing, AI, and ML in mental health care.	Passive sensing tools (e.g., smartphones, wearables) and AI/ML models used for behavior tracking and mental health monitoring.	Mental health professionals (e.g., clinicians, nurses, psychiatrists) evaluating digital interventions and their implications.	Themes including utility, empowerment, risks, workload impact, data protection, and clinician-user relationships.	Clinicians support AI but stress the need for ethical guidelines, risk management, and co-development with users.
⁵⁵	Rahsepar Meadi et al. (2025)	Scoping review	To provide a comprehensive overview of ethical considerations around conversational AI (CAI) in mental health therapy.	Conversational AI (CAI), including psychotherapeutic chatbots and AI-driven therapeutic agents in clinical and nonclinical settings.	Mental health service users using CAI for support or therapy; includes discussions on clinical and nonclinical settings.	Ten ethical themes: safety, explicability, accountability, empathy, justice, anthropomorphization, autonomy, effectiveness, privacy, job security.	Major ethical concerns include trust, data privacy, responsibility, potential harm, and human replacement risks.
⁵⁶	Scherbakov et al. (2025)	Scoping review	To summarize the use of NLP in mental health research with focus on text dataset types and the integration of social determinants of health (SDOH).	Natural language processing (NLP), large language models (LLMs), social media and clinical notes analysis, sentiment mining, and EHR text mining.	Mental health patients; studies focus on depression, suicide, and psychiatric disorders using text data.	Dataset types, frequency of SDOH use, accessibility of datasets, NLP method deployment, and study coverage of demographic data.	Clinical and social media texts dominate NLP in mental health; SDOH is underutilized; many datasets lack transparency or access.
⁵⁷	Cecula et al. (2021)	Narrative literature review	To explore how AI can be used to improve patient flow in mental health inpatient units, and identify gaps for future research.	AI applications for diagnosis, prognosis, and treatment using ML, SVM, clustering, neuroimaging, and NLP in psychiatric inpatient units.	Mental health inpatients; studies include patients with psychosis, depression, and schizophrenia.	Themes of diagnosis, prognosis, treatment, and patient flow with performance metrics like AUC, accuracy, sensitivity, and AI integration issues.	AI holds potential to enhance patient flow in inpatient settings; however, few studies address this directly; privacy and operational integration remain gaps.
⁵⁸	Bhatt et al. (2022)	Scoping review	To synthesize current research on AI-powered mobile health (mHealth) applications for managing diseases and promoting health.	AI models using wearable sensors, deep learning, federated learning, remote monitoring systems, predictive diagnostics, and disease management tools.	General population including individuals with depression, suicidal tendencies, and chronic conditions using wearable health devices.	Prevalence of AIM models in healthcare delivery, application domains, health outcomes, public vs. proprietary dataset use, and gaps in chronic mental health research.	AIM models are growing in use but underrepresent chronic mental health care and rely heavily on private datasets; call for better transparency and broader deployment.
⁵⁹	Ding et al. (2024)	Systematic review and meta-analysis	To conduct a meta-analysis assessing the diagnostic accuracy of deep learning (DL) techniques for ASD classification in children.	Deep learning (CNN, DNN, RNN, BLSTM), applied to facial imaging, fMRI, clinical data for ASD prediction.	Children diagnosed with ASD across different countries using datasets like ABIDE and Kaggle.	Sensitivity (0.95), specificity (0.93), AUC (0.98) for DL models in ASD detection; dataset performance variation analyzed.	DL shows excellent diagnostic accuracy for ASD; however, heterogeneity and clinical application need further validation.
⁶⁰	Su et al. (2020)	Scoping review	To review existing research on DL applications in mental health outcome research, categorized by data types and clinical relevance.	DL models (DFNN, CNN, RNN, Autoencoder) applied to neuroimaging, genetic, vocal/visual expression, and social media data.	Patients with mental illnesses (e.g., ADHD, depression, schizophrenia, ASD), including adolescents and adults.	Application performance categorized by data type; highlights challenges in model interpretability and real-world application.	DL has promising application across diverse mental health domains; interpretability and ethics remain challenges.
⁶¹	Islam et al. (2025)	Systematic review	To assess AI-based risk assessment tools for sexual, reproductive, and mental health, and identify challenges and future research directions.	AI-based triage tools, symptom checkers, risk prediction models using clinical, demographic, and online data.	General population with focus on users in LMICs seeking SRMH and mental health services.	Effectiveness of AI tools in healthcare outcomes, disparities in adoption, and ethical gaps in real-world implementation.	AI-based tools improve SRMH and mental health delivery, but require better dataset diversity, ethical practices, and local applicability.
⁶²	Alabdulkareem et al. (2022)	Systematic review	To review and classify studies on Robot-Assisted Autism Therapy (RAAT) for children, assessing robot types, therapeutic targets, and developmental outcomes.	Socially assistive robots (e.g., Nao, Kasper) for social, emotional, cognitive, and communication development in therapy for children with ASD.	Children with autism spectrum disorder, primarily between 3–12 years old, participating in RAAT programs.	Improvements in communication, imitation, eye contact, emotional expression, social skills; robot type and setting effects.	RAAT shows promise in enhancing emotional, social, and cognitive skills in children with ASD, but standardization and long-term studies are lacking.
⁶³	Solek et al. (2025)	Systematic review	To examine AI applications in ASD diagnosis among pediatric populations, focusing on accuracy and data modalities used for early intervention.	AI techniques (ML, DL, SVM, LSTM, CNN) using multimodal data: facial imaging, behavior, microbiome, genetics, motor function, and neuroimaging.	Children aged 0–18 with ASD and typically developing controls; focus on diagnostic classification using AI.	Accuracy, sensitivity, and specificity of AI models; predictive strength and early detection performance across data types.	AI demonstrates strong diagnostic capabilities for ASD, especially when using combined data modalities; ethics and clinical integration remain concerns.
⁶⁴	Kohli et al. (2022)	Scoping review	To analyze the role and evolution of intelligent technologies in early ASD detection, focusing on data sources, demographics, and detection outcomes.	ML and DL models applied to behavioral, eye-tracking, facial expression, postural, auditory, and EHR data for early ASD screening in children under 6.	Children under 6 years of age, especially at risk of ASD or with developmental concerns.	Technological readiness, multimodal data utility, ethical limitations, model generalizability, and diagnostic accuracy.	Intelligent tech enables earlier ASD detection, but research must improve data quality, design protocols, and real-world application to validate tools.
⁶⁵	Kolding et al. (2025)	Systematic review	To systematically review the literature on generative AI applications in psychiatry and mental health, exploring tools like ChatGPT and their clinical and ethical implications.	Generative AI tools (e.g., ChatGPT, DALLÂ·E, Bard, Claude.ai, Replika), used for diagnosis, psychotherapy, psychoeducation, and patient support in psychiatry.	Individuals with mental health issues (e.g., depression, schizophrenia, substance use); general population; clinicians and educators.	Accuracy of knowledge responses, psychoeducational content, clinical tool performance, ethical risks (e.g., bias, misinformation), and AI's emotional processing.	Generative AI shows high potential in psychiatric settings but suffers from ethical and methodological limitations; most studies are exploratory with limited clinical validation.
⁶⁶	Li et al. (2023)	Systematic review and meta-analysis	To evaluate the effectiveness of AI-based conversational agents (CAs) on mental health and well-being, and identify key factors influencing their success and user experiences.	AI-based conversational agents (e.g., Woebot, Tess, XiaoE, Wysa), using NLP, ML, and emotional algorithms, delivered via smartphone apps, messengers, and voice interfaces.	Clinical and subclinical populations, adolescents, adults, elderly; studies included both diagnosed patients and general users.	Symptom reduction (depression, distress), user engagement, psychological well-being, therapeutic alliance, and safety concerns of AI interventions.	AI-based CAs significantly reduce depression and distress symptoms, especially when multimodal and integrated into everyday platforms; long-term effects remain underexplored.

HER: human emotion recognition; ASD: autism spectrum disorder; ML: machine learning; SVM: support vector machine; AI: artificial intelligence; CNN: convolutional neural network; IoT: internet of things; OCD: obsessive–compulsive disorder; RF: random forest; AUC: area under the curve; MRI: magnetic resonance imaging; fMRI: functional magnetic resonance imaging; HOG: histogram of oriented gradients; LSTM: long short-term memory; R-CNN: region-based convolutional neural network; DSS: decision support systems; NLP: natural language processing; EHR: electronic health records; KNN: K-nearest neighbors; RNN: recurrent neural network; SAR: socially assistive robots; CAI: conversational artificial intelligence; SDOH: social determinants of health; mHealth: mobile health; DL: deep learning; DNN: deep neural network; BLSTM: bidirectional long short-term memory; DFNN: deep feedforward neural network; ADHD: attention-deficit/hyperactivity disorder; SRMH: sexual, reproductive, and mental health; LMICs: low- and middle-income countries; RAAT: robot-assisted autism therapy; LLMs: large language models; CA: conversational agents; CAs: conversational agents.

Thematic analysis and data synthesis

Given the qualitative emphasis of this umbrella review, thematic analysis served as the primary methodological framework for synthesizing the extracted data. The thematic analysis facilitated systematic identification, analysis, and interpretation of recurring patterns, themes, and critical insights within the literature. Reviewers initiated this process by deeply immersing themselves in the extracted data, achieving a comprehensive understanding of initial patterns and preliminary themes. Following initial data familiarization, a structured coding process was implemented. During this stage, meaningful segments of data were carefully coded, reflecting their direct relevance to the research questions guiding the umbrella review. Generated codes were subsequently organized into broader thematic categories and detailed sub-themes through iterative review and refinement, ensuring accurate representation of data nuances. This rigorous and iterative approach provided clearly defined thematic constructs, enabling a coherent and meaningful synthesis of the evidence. Core themes identified through thematic analysis included the effectiveness and therapeutic benefits of AI-driven art therapy, specific impacts on autism spectrum disorders, mental health improvements, enhancements in emotional well-being, and critical implementation aspects, including technological considerations, user experiences, and accessibility factors. Thematic maps visually illustrate the interconnections among identified themes and sub-themes, offering deeper insights into the mechanisms and contexts through which AI-driven art therapy achieves therapeutic outcomes.

Reporting standards

This umbrella review rigorously adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for comprehensive reporting.⁶⁷ A detailed flow diagram was provided, outlining each step of the literature screening and selection process—from initial database searches through various screening stages to the final set of included reviews. As illustrated in Figure 1, the search initially yielded 139 records, of which duplicates, and irrelevant articles were removed. After screening and exclusions due to methodological flaws or insufficient data, 30 reviews were included in the final synthesis, ensuring a rigorous and transparent selection process. Compliance with established reporting standards ensured thorough, transparent, and reproducible documentation of methodological and analytical procedures. The comprehensive methodological framework and systematic rigor employed throughout this umbrella review allowed for a detailed synthesis of existing literature on AI-driven art therapy. By incorporating systematic and scoping reviews, this approach enabled a broad yet detailed understanding of the current evidence landscape, contributing valuable insights and identifying pivotal knowledge gaps relevant for future research, clinical practice, and therapeutic innovations addressing autism, mental health, and emotional well-being.

Figure 1.

PRISMA flow diagram.

Results

This umbrella review took a deep dive into how AI-powered art therapy is being used to support individuals with ASD, mental health challenges, and issues related to emotional well-being. By analyzing themes across various review articles, the study uncovered meaningful insights into how these technologies work, what benefits they offer, and what obstacles remain when it comes to practical application. As shown in Figure 2, AI-driven art therapy supports outcomes such as emotion regulation, anxiety and stress reduction, enhanced coping skills, improved social interaction, accessible interfaces, and tailored therapeutic sessions, highlighting its multidimensional impact on mental health and well-being.

Figure 2.

AI-driven art therapy outcomes.

Overview of included reviews

A total of 30 reviews were included in this umbrella review.^37–66 Of these, the majority comprised systematic reviews (40.0%) and scoping reviews (36.7%), reflecting a balanced use of comprehensive synthesis and exploratory mapping methods. Additionally, systematic reviews and meta-analyses accounted for 10%, followed by integrative reviews (6.7%), which offered broader interpretive insights across diverse methodologies. Other designs included systematic reviews with meta-synthesis (3.3%), and narrative literature reviews (3.3%), each contributing distinct methodological perspectives.

This diversity in review types underscores the interdisciplinary and evolving nature of research at the intersection of AI and art therapy. The frequent use of systematic and scoping reviews demonstrates a dual emphasis on rigorously evaluating existing interventions and identifying emerging themes within the field. Meanwhile, meta-analytical and integrative designs added depth by highlighting effect sizes and synthesizing findings across different study contexts. The inclusion of a range of methodologies enriched the overall synthesis process and allowed for a nuanced understanding of AI-driven art therapy's applications across autism, mental health, and emotional well-being.

Effectiveness and therapeutic benefits of AI-driven art therapy

A significant theme highlighted across reviewed studies was the therapeutic effectiveness of AI-driven art therapy. Reviews consistently reported enhanced communication skills, improved social interactions, and greater emotional expression among ASD populations.⁴¹ AI technologies, particularly ML algorithms, proved notably effective in customizing therapy sessions according to individual user needs, resulting in increased engagement and positive therapeutic outcomes.³⁹ Enhanced cognitive-emotional processing was frequently noted, attributed primarily to personalized AI-driven therapeutic modules tailored for specific therapeutic interventions.³⁸

In mental health contexts, AI-driven art therapy demonstrated substantial therapeutic benefits for individuals experiencing anxiety, depression, and PTSD.⁵⁴ Findings indicated that participation in AI-supported artistic activities significantly mitigated symptoms related to anxiety and depressive episodes.⁵⁷ Interactive, adaptive art therapy platforms leveraging AI were particularly effective at promoting emotional regulation, stabilizing mood, and enhancing overall psychological resilience.⁵⁶ These platforms enabled users to externalize inner emotions through art while receiving real-time AI-generated support, contributing to a more dynamic and empathetic therapeutic process.⁴⁸ Some reviews also pointed to the preventive role of AI-driven art therapy in at-risk populations.^44,66 By encouraging early emotional expression and offering adaptive coping strategies, these systems helped reduce the escalation of mental health symptoms in youth and vulnerable groups.^47,55 Moreover, the aesthetic and creative engagement fostered by AI-powered art tools contributed to a sense of accomplishment and purpose among users, further reinforcing positive behavioral changes.⁵⁸

Specific impacts on autism spectrum disorders

Within the ASD context, AI-driven art therapies were particularly impactful in addressing primary symptoms such as social deficits, repetitive behaviors, and communication barriers.³⁷ Studies highlighted AI technologies’ capability to monitor user behavior in real-time, providing tailored feedback and adaptable content designed explicitly for autistic individuals’ developmental needs.^40,52 The integration of virtual reality (VR) and augmented reality (AR) with AI was notably transformative,⁵³ creating immersive environments where participants could safely navigate emotional and social scenarios without typical anxiety associated with real-world interactions.⁴⁶ de Belen et al. (2020) emphasized improvements in both verbal and non-verbal communication skills, highlighting AI-driven platforms’ ability to interpret subtle cues such as eye gaze, facial expressions, tone of voice, and gesture recognition rather than remaining limited to general behavioral descriptions.⁴³ For example, some platforms used computer vision and deep learning algorithms to detect and respond to facial expressions or gaze-shifting patterns, while others employed natural language processing to evaluate speech rhythm and prosody.³⁷

AI-driven interventions’ adaptability enabled continuous and incremental adjustments, fostering prolonged engagement and sustained skill development.⁶³ Additionally, these systems facilitated structured art-based interactions in which individuals could engage without sensory overload or fear of misinterpretation,⁵⁹ which are common challenges in traditional therapy settings for those on the spectrum.⁶² Examples of such structured activities included AI-guided digital drawing tasks with real-time feedback, VR and AR tools that created immersive environments to support emotion recognition, and music composition programs that adjusted rhythm and melody.³⁷ As many individuals with autism thrive in structured environments, AI systems that followed predictable routines and reinforced positive behaviors were particularly effective.⁶⁴ The capacity for repetition and individual pacing also allowed for mastery and confidence-building in a safe, non-judgmental space.⁴⁰ In support of these impacts, a systematic review analyzing 944 studies and including 40 eligible articles reported that AI- and ML-based diagnostic models for ASD often reached very high performance, with some deep learning approaches achieving accuracy above 99% in distinguishing autistic from non-autistic individuals.⁴⁶

Improvements in mental health

The umbrella review found robust evidence supporting AI-driven art therapy's efficacy in mental health management. These therapies facilitated emotional release and introspection, offering safe environments for emotional exploration, crucial for mental health recovery.⁴⁴ Outcomes included decreased stress levels, diminished anxiety, and improved coping mechanisms, significantly enhancing patients’ mental health and quality of life.⁴⁸ ML-driven systems—such as those employing support vector machines and convolutional neural networks—provided customized therapeutic content by adapting to user preferences, psychological status, and therapeutic progression, as reported in systematic reviews analyzing digital interventions for anxiety and depression.^51,54 These reviews highlighted that personalization was derived from training on behavioral and clinical datasets, which improved both relevance and effectiveness. Personalized therapeutic experiences created by these adaptive platforms were essential for increasing patient adherence and achieving lasting therapeutic outcomes.⁴² Moreover, several reviews cited the potential for AI to reduce the stigma associated with seeking mental health support.^45,60 By delivering therapy through digital platforms in non-clinical environments, AI-driven art therapy allowed users to engage with mental health interventions discreetly and at their own pace.⁵⁷ This flexibility was especially beneficial for individuals hesitant to participate in traditional therapy due to privacy concerns, social anxiety, or cultural barriers.⁶⁵

Enhancements in emotional well-being

A prominent finding across reviews was AI-driven art therapy's significant role in enhancing emotional awareness, regulation, and resilience.⁵⁰ Participants engaged in AI-supported creative activities—such as AI-guided digital drawing tasks, interactive painting programs, and robot-assisted autism therapy (RAAT)—showed considerable improvements in self-awareness, emotional expression, and emotional regulation capabilities.⁴⁸ AI-driven platforms offered tailored emotional support through adaptive feedback mechanisms, significantly improving user experiences and therapeutic results.⁶⁰ Interactive AI features allowed real-time emotional state monitoring, providing instant, constructive feedback crucial for emotional growth.⁴⁷ Such capabilities effectively promoted emotional literacy, fostered greater resilience, and empowered users with strategies for managing emotional challenges.⁶² In addition to therapeutic outcomes, emotional well-being improvements were tied to participants’ sense of autonomy and control.⁶⁶RAAT shows promise in enhancing emotional, social, and cognitive skills in children with ASD, but standardization and long-term studies are lacking.⁶² For many, the non-verbal nature of art served as a bridge to emotional insight, which was further enriched by AI's ability to provide context-sensitive prompts, affirmations, or guided questions to facilitate deeper reflection.⁴⁹

Technological considerations and implementation barriers

Despite notable therapeutic potential, reviews consistently identified technological challenges as significant barriers to implementation.⁶¹ Issues included the complexity of AI algorithms, substantial initial setup costs, and ongoing technological maintenance and upgrades requirements.⁵⁰ Privacy concerns, data security issues, and ethical considerations surrounding AI surveillance emerged as critical factors necessitating rigorous ethical guidelines and user consent protocols to prevent potential misuse or data breaches.⁴¹ Further technical barriers included limited digital literacy among users and practitioners, complicating the widespread adoption of AI-driven art therapy.⁶¹ Successful implementation required comprehensive training, user-friendly technological interfaces, and ongoing technical support to ensure sustainability and efficacy.⁶⁶ Scalability and interoperability were also noted as significant concerns. Integrating AI-driven therapy systems into existing clinical workflows often proved challenging due to incompatibilities with electronic health records, insufficient regulatory guidance, or institutional resistance.⁴² Moreover, the lack of standardized protocols for AI integration in therapeutic settings limited widespread adoption, particularly in under-resourced areas.⁴⁸

User experiences and accessibility factors

User experience emerged as a crucial determinant of AI-driven art therapy's success.^37,39 Reviews underscored the importance of user-centered design in significantly influencing user engagement and therapeutic outcomes.^42,46 Factors such as intuitive interfaces, ease of interaction, and personalized adaptive feedback considerably improved user acceptance and participation.⁴⁸

Accessibility was another critical theme, with variations noted depending on demographics, socioeconomic status, and geographic location.⁶² The digital divide emerged as a significant barrier, emphasizing the need for inclusive technological designs that accommodate diverse user requirements and foster equitable access to therapeutic resources.⁵⁶ AI-driven platforms that prioritized inclusivity notably enhanced therapeutic effectiveness and user satisfaction.⁴⁴ Several reviews proposed strategies to enhance accessibility, including offline-compatible applications, culturally adapted content, and multilingual interfaces.^42,49 Such innovations were essential for ensuring broader adoption, especially in marginalized communities or global regions where internet connectivity and digital infrastructure remain limited.⁶⁴ Additionally, positive user experiences were linked to the perceived sense of anonymity and safety that AI-driven art therapy platforms could offer.³⁹ This aspect was particularly relevant for individuals recovering from trauma or facing social stigmatization, who found solace in engaging with technology-based art therapies without fear of judgment.⁵⁵

Discussion

The findings of this umbrella review underscore the dynamic intersection of technology and therapeutic practice, revealing how AI-driven art therapy is reshaping emotional and psychological support for diverse populations. Central to this innovation is the capacity of AI technologies to enhance traditional art therapy by delivering highly personalized, adaptive, and emotionally resonant interventions.⁶⁸ This personalization is particularly transformative for individuals with ASD, who often struggle with verbal communication, sensory sensitivities, and social interactions.⁶⁹ AI-supported art therapy platforms created safe and structured environments where users could express complex emotions visually and nonverbally.⁷⁰ These digital platforms, powered by ML and pattern recognition, analyzed user input and responded with tailored prompts and context-sensitive modifications, enabling continuous engagement that traditional therapies might not sustain.^71,72 Many included studies pointed to meaningful improvements in users’ social behavior, emotional articulation, and cognitive processing, attributing these outcomes to AI's ability to respond to individual emotional needs in real time.⁷³ The structured predictability of AI tools also aligned well with the needs of autistic individuals, fostering therapeutic consistency and comfort without overwhelming stimuli.⁷⁴

Parallel to these findings, the application of AI-driven art therapy in mental health contexts emerged as a significant contributor to emotional stabilization and symptom reduction in individuals facing depression, anxiety, and trauma-related disorders. Several reviews specifically reported that AI-enhanced platforms—such as generative art systems and emotion-recognition tools integrated within VR-based therapy environments—reduced depressive symptoms and supported coping strategies for individuals with PTSD and anxiety. These tools offered an interactive space where users could safely externalize internal struggles, track emotional patterns, and reflect through artistic creation.^75–77 The integration of intelligent feedback systems allowed users to receive supportive and constructive responses tailored to their current emotional state, fostering a deeper therapeutic connection.⁷⁸ Importantly, AI platforms helped bridge the gap for individuals who experience stigma or discomfort around traditional therapy settings by offering an alternative that was both discreet and non-clinical.⁷⁹ The empowering nature of these tools—granting users control over their self-expression and pace of engagement—added to the therapeutic efficacy, reinforcing users’ autonomy and agency in managing their mental well-being.^80,81 Additionally, the positive psychological reinforcement provided through AI-driven art activities often instilled a sense of accomplishment and personal validation, key components in fostering long-term emotional strength.⁸²

Despite the evident therapeutic benefits, several implementation challenges and limitations emerged across the included studies, drawing attention to the practical and ethical complexities of embedding AI into therapeutic frameworks. One of the most frequently cited concerns was the technological sophistication required to develop and maintain these platforms.⁸³ Many AI systems used in art therapy rely on advanced algorithms, immersive environments, or emotion-sensing technologies,⁸⁴ which demand significant technical infrastructure, continuous updates, and user training.⁸⁵ These requirements pose significant challenges for widespread adoption, especially in low-resource settings.⁸⁶ Moreover, ethical issues surrounding data privacy and consent were recurrent themes in the literature.⁸⁷ Since AI platforms often gather sensitive psychological and behavioral data, the risk of data breaches, misuse, or lack of transparency in algorithmic decisions raises valid concerns.⁸⁸ The absence of robust ethical protocols and regulatory standards for AI in therapy further complicates these concerns,⁸⁹ leaving practitioners and users without clear guidelines for safeguarding user data and ensuring equitable outcomes.⁹⁰ Several studies have also emphasized broader ethical and political issues, such as epistemic inequality and structural digital divides, which determine whose cultural frameworks shape AI design and who has meaningful access to these technologies.^91–93 Another layer of complexity is the presence of algorithmic bias or insufficient cultural sensitivity in AI-generated feedback, which may inadvertently lead to misinterpretation or reduced therapeutic relevance for individuals from diverse backgrounds.⁹⁴ These systemic issues reflect broader tensions between technological advancement and ethical responsibility, particularly when AI is applied to emotionally vulnerable populations.⁹⁵

In addition to these barriers, user experience and accessibility emerged as essential factors influencing the effectiveness and equity of AI-driven art therapy.⁹⁶ The degree to which platforms are intuitively designed and culturally attuned significantly impacted user engagement and therapeutic success.⁹⁷ A study by Connors et al. (2025) highlighted that platform with simple navigation, real-time feedback, and customizable features tended to produce more sustained participation and better therapeutic outcomes.⁹⁸ Conversely, platforms that lacked user-centered design or failed to accommodate neurodivergent needs—such as overstimulating visuals or confusing interfaces—often discouraged engagement.⁹⁹ Accessibility was also shaped by socioeconomic and geographical factors, with the digital divide limiting access for those in under-resourced communities or areas with poor internet connectivity.¹⁰⁰ This divide posed a threat to equitable therapeutic access, as individuals most in need of support may also be least likely to benefit from these innovations.¹⁰¹ Furthermore, users’ digital literacy and comfort with technology significantly influenced their ability to engage with AI-based therapy.¹⁰² While younger populations or those familiar with digital tools might navigate these platforms with ease, older adults or individuals unfamiliar with technology may face challenges that reduce the platforms’ therapeutic potential.¹⁰³ Nonetheless, when effectively implemented, AI-driven art therapy platforms created unique and empowering therapeutic spaces, particularly appreciated for their anonymity, autonomy,¹⁰⁴ and emotional safety—allowing users to process trauma, manage stress, and enhance well-being in ways that felt personal and non-threatening.¹⁰⁵

Implications of the study

The findings from this umbrella review hold substantial implications for clinical practice, digital health innovation, and the broader landscape of psychological support. AI-driven art therapy, through its adaptive and emotionally responsive design, presents a powerful adjunct to conventional therapeutic methods. For individuals with ASD, these technologies offer structured, predictable environments that foster communication, social engagement, and emotional expression. The real-time feedback and personalization capabilities of AI can significantly improve therapeutic outcomes by accommodating unique user needs and preferences, particularly for neurodivergent individuals who may find traditional therapy settings overwhelming or ineffective. These tools not only bridge communication gaps but also support skill development through repetitive, non-judgmental, and interactive sessions—an advantage especially relevant in managing ASD-related challenges.

Moreover, in mental health contexts, AI-enhanced art therapy platforms provide a novel, low-barrier entry point for individuals seeking psychological support. The opportunity to engage in therapy from the comfort of one's own space, combined with anonymity and emotional safety, reduces stigma and increases accessibility—particularly for those reluctant or unable to access conventional services. For populations dealing with anxiety, depression, or trauma, AI-based systems offer real-time emotional support, enhance emotional literacy, and foster resilience. Importantly, these platforms promote self-efficacy and autonomy by allowing users to navigate therapy at their own pace. However, for these promising benefits to be fully realized, ethical and technical challenges such as data security, algorithmic transparency, and digital inclusivity should be addressed. The implications extend beyond individual therapy to systemic innovation in mental health delivery, highlighting the need for responsible integration of AI in therapeutic practice that ensures both effectiveness and equity.

Limitations

This umbrella review provides valuable insights but is not without limitations. The synthesis depended entirely on previously published systematic and scoping reviews, which means that the strength of the conclusions is tied to the quality and scope of those original works. Reliance on secondary sources may also introduce the risk of bias, since studies reporting positive outcomes are often more likely to appear in peer-reviewed literature. Considerable variability across included studies, such as differences in populations, intervention designs, and AI applications, further complicates the ability to generate consistent conclusions or make precise comparisons. It is also worth noting that while this review included both pediatric and adult populations, therapeutic needs may differ between these groups, and future research should further examine age-specific effectiveness to ensure that interventions are developmentally appropriate and optimally tailored.

The rapidly evolving nature of AI technologies presents another challenge, as many studies examined early-stage or pilot programs with uncertain scalability, reproducibility, and long-term sustainability. In addition, some interventions were commercially developed, and studies with industry involvement may carry conflicts of interest that influence interpretation of results. Ethical concerns such as data privacy, informed consent, and algorithmic transparency were also insufficiently addressed across the literature. These limitations suggest that while AI-driven art therapy is promising, the findings should be interpreted carefully and situated within broader ethical and practical contexts.

Recommendations for future research

While this umbrella review has consolidated valuable insights into the potential of AI-driven art therapy, future research should aim to fill several critical gaps to advance this emerging field. Longitudinal studies are essential to evaluate the sustained therapeutic effects and long-term user engagement with AI-enhanced platforms. It is also necessary to compare these technologies with traditional art therapy and other digital mental health interventions to identify their relative strengths, limitations, and ideal user profiles. Furthermore, future studies should prioritize inclusivity by incorporating diverse cultural, linguistic, and socioeconomic contexts to ensure global applicability. Research exploring accessibility for individuals with disabilities, older adults, and those in remote or underserved areas is particularly important in bridging the digital divide. In parallel, investigations into the ethical dimensions of AI uses—such as data privacy, informed consent, and algorithmic fairness—should be rigorously pursued to build trust and transparency within therapeutic frameworks. Another key area involves participatory design approaches that engage end-users, including autistic individuals and those with mental health concerns, in co-creating platforms that are intuitive, emotionally attuned, and clinically appropriate. Such collaborative development can improve usability, foster greater emotional resonance, and reduce potential harm or misunderstanding. By addressing these dimensions, future research can contribute to the safe, effective, and inclusive expansion of AI-driven art therapy within modern mental health care.

Conclusion

This umbrella review highlights that AI-driven art therapy offers meaningful therapeutic value across autism, mental health, and emotional well-being, yet its promise is tempered by notable gaps. Evidence indicates consistent improvements in communication, emotional regulation, and self-expression, particularly for individuals with ASD, and reductions in anxiety, depression, and trauma-related symptoms in broader mental health contexts. However, the field remains fragmented, with an overreliance on supervised machine learning and facial-cue–based interactions, while deep learning applications and culturally inclusive datasets are underexplored. Implementation is further constrained by ethical concerns, data privacy issues, and accessibility barriers, which limit scalability and equitable adoption. These findings suggest that AI-driven art therapy is best positioned as an adjunct rather than a replacement for human-led care, offering unique opportunities for personalization and engagement. Future progress will depend on context-sensitive system design, stronger ethical frameworks, and inclusive research that integrates user perspectives across diverse populations.

Footnotes

Acknowledgments

The authors are deeply grateful to the Miyan Research Institute, International University of Business Agriculture and Technology, Dhaka, Bangladesh.

ORCID iD

Moustaq Karim Khan Rony

Ethical approval

Our study did not require an ethical board approval because it did not contain human or animal trials.

Contributorship

Conceptualization: Sumaiya Yeasmin, Sanchita Saha, Moustaq Karim Khan Rony, Mst Masuma Akter Semi, Srabani Das, Rayhan Khan. Data curation: Mst Masuma Akter Semi, Rukshanda Rahman, Afia Fairooz Tasnim, Arif Hosen. Formal analysis: Sumaiya Yeasmin, Rayhan Khan. Investigation: Moustaq Karim Khan Rony, Sanchita Saha, Afia Fairooz Tasnim, Mst Masuma Akter Semi. Methodology: Moustaq Karim Khan Rony, Sumaiya Yeasmin, Srabani Das. Project administration: Sumaiya Yeasmin, Afia Fairooz Tasnim, Sanchita Saha. Resources: Rukshanda Rahman, Arif Hosen, Srabani Das.

Software: Moustaq Karim Khan Rony, Sanchita Saha, Rayhan Khan. Supervision: Moustaq Karim Khan Rony, Sumaiya Yeasmin, Mst Masuma Akter Semi. Validation: Afia Fairooz Tasnim, Rukshanda Rahman. Visualization: Sumaiya Yeasmin, Arif Hosen. Writing—original draft: Sumaiya Yeasmin, Srabani Das, Moustaq Karim Khan Rony, Mst Masuma Akter Semi, Rukshanda Rahman, Rayhan Khan. Writing—review and editing: Sanchita Saha, Moustaq Karim Khan Rony, Sumaiya Yeasmin, Arif Hosen. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work.

Funding

There was no external fund taken for this current research.

Declaration of conflict of interest

The authors have no competing interest at all.

References

Kalmady

Greiner

Agrawal

, et al. Towards artificial intelligence in mental health by improving schizophrenia prediction with multiple brain parcellation ensemble-learning. npj Schizophr 2019; 5: 2.

Jethwa

Lote

. A review on how AI helps in treating mental illness2024 2nd DMIHER international conference on artificial intelligence in healthcare, education and industry (IDICAIEI). India: IEEE, 2024, pp.1–6. doi:10.1109/IDICAIEI61867.2024.10842844.

Vial

Almon

. Artificial intelligence in mental health therapy for children and adolescents. JAMA Pediatr 2023; 177: 1251.

Fiske

Henningsen

Buyx

. Your robot therapist will see you now: ethical implications of embodied artificial intelligence in psychiatry, psychology, and psychotherapy. J Med Internet Res 2019; 21: e13216.

Atlam

Rokaya

Masud

, et al. Explainable artificial intelligence systems for predicting mental health problems in autistics. Alexandria Eng J 2025; 117: 376–390.

Shahamiri

Thabtah

. Autism AI: a new autism screening system based on artificial intelligence. Cogn Comput 2020; 12: 766–777.

Chaddad

, et al. Can autism be diagnosed with artificial intelligence? A narrative review. Diagnostics 2021; 11: 2032.

Thakkar

Gupta

De Sousa

. Artificial intelligence in positive mental health: a narrative review. Front Digit Health 2024; 6: 1280235.

Abd-alrazaq

Alhuwail

Schneider

, et al. The performance of artificial intelligence-driven technologies in diagnosing mental disorders: an umbrella review. npj Digit Med 2022; 5: 87.

10.

Ray

Bhardwaj

Malik

, et al. Artificial intelligence and psychiatry: an overview. Asian J Psychiatr 2022; 70: 103021.

11.

Mengi

Malhotra

. Artificial intelligence based techniques for the detection of socio-behavioral disorders: a systematic review. Arch Computat Methods Eng 2022; 29: 2811–2855.

12.

Olawade

Wada

Odetayo

, et al. Enhancing mental health with artificial intelligence: current trends and future prospects. Journal of Medicine, Surgery, and Public Health 2024; 3: 100099.

13.

Pham

Nabizadeh

Selek

. Artificial intelligence and chatbots in psychiatry. Psychiatr Q 2022; 93: 249–253.

14.

Lee

Torous

De Choudhury

, et al. Artificial intelligence for mental health care: clinical applications, barriers, facilitators, and artificial wisdom. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 2021; 6: 856–864.

15.

Graham

Depp

Lee

, et al. Artificial intelligence for mental health and mental illnesses: an overview. Curr Psychiatry Rep 2019; 21: 16.

16.

Straw

Callison-Burch

. Artificial intelligence in mental health and the biases of language based models. PLoS ONE 2020; 15: e0240376. Danforth CM, ed.

17.

Tornero-Costa

Martinez-Millana

Azzopardi-Muscat

, et al. Methodological and quality flaws in the use of artificial intelligence in mental health research: systematic review. JMIR Ment Health 2023; 10: e42045.

18.

D’Alfonso

Santesteban-Echarri

Rice

, et al. Artificial intelligence-assisted online social therapy for youth mental health. Front Psychol 2017; 8: 96.

19.

Denecke

Abd-Alrazaq

Househ

. Artificial intelligence for chatbots in mental health: opportunities and challenges. In: Househ

Borycki

Kushniruk

(eds) Multiple perspectives on artificial intelligence in healthcare. Lecture notes in bioengineering. Switzerland: Springer International Publishing, 2021, pp.115–128. doi:10.1007/978-3-030-67303-1_10.

20.

Liu

Ning

Zhang

, et al. Estimation of the global disease burden of depression and anxiety between 1990 and 2044: an analysis of the global burden of disease study 2019. Healthcare 2024; 12: 1721.

21.

World Health Organization . Post-traumatic stress disorder. Switzerland: World Health Organization, 2024. Retrieved from https://www.who.int/news-room/fact-sheets/detail/post-traumatic-stress-disorder.

22.

Alhuwaydi

. Exploring the role of artificial intelligence in mental healthcare: current trends and future directions – A narrative review for a comprehensive insight. RMHP 2024; 17: 1339–1348.

23.

Jin

Xie

, et al. Artificial intelligence in mental healthcare: an overview and future perspectives. Br J Radiol 2023; 96: 20230213.

24.

Mody

. Mental health monitoring system using artificial intelligence: a review2019 IEEE 5th international conference for convergence in technology (I2CT). India: IEEE, 2019, pp.1–6. doi:10.1109/I2CT45611.2019.9033652.

25.

King

Nanda

Stoddard

, et al. An Introduction to generative artificial intelligence in mental health care: considerations and guidance. Curr Psychiatry Rep 2023; 25: 839–846.

26.

Joyce

Kormilitzin

Smith

, et al. Explainable artificial intelligence for mental health through transparency and interpretability for understandability. npj Digit Med 2023; 6: 6.

27.

Torous

Blease

. Generative artificial intelligence in mental health care: potential benefits and current challenges. World Psychiatry 2024; 23: –2.

28.

Nag

Das

Sil

, et al. Application of artificial intelligence in mental health. In: Abraham

Pllana

Casalino

, etal (eds) Intelligent systems design and applications. Vol 646. Lecture notes in networks and systems. Switzerland: Springer Nature Switzerland, 2023, pp.128–141. doi:10.1007/978-3-031-27440-4_13.

29.

Luo

Zhang

, et al. Emergence of artificial intelligence art therapies (AIATs) in mental health care: a systematic review. Int J Mental Health Nurs 2024; 33: 1743–1760.

30.

Shojaei

Osorio Torres

, et al. Insights from art therapists on using AI-generated art in art therapy: mixed methods study. JMIR Form Res 2024; 8: e63038–e63038.

31.

Shi

Wang

, et al. Harmonizing emotions: an AI-driven sound therapy system design for enhancing mental health of older adults. In: Degen

Ntoa

(eds) Artificial intelligence in HCI. Lecture notes in computer science. 14736. Switzerland: Springer Nature Switzerland, 2024, pp.439–455. doi:10.1007/978-3-031-60615-1_30.

32.

Zhang

Lin

Luo

, et al. Co-creation with AI: a painting therapy program aimed at ameliorating attention deficits in children with ADHD. Applied Neuropsychology: Child 2024: 1–10. Published online December 21, 2024. doi:10.1080/21622965.2024.2442100

33.

Panda

Abraham

Gopi

etal (eds) Computational Intelligence for Oncology and Neurological Disorders: Current Practices and Future Directions. Florida: CRC Press, 2024.

34.

Fiske

Henningsen

Buyx

. The implications of embodied artificial intelligence in mental healthcare for digital wellbeing. In: Burr

Floridi

(eds) Ethics of digital well-being. Philosophical studies series. 140. Switzerland: Springer International Publishing, 2020, pp.207–219. doi:10.1007/978-3-030-50585-1_10.

35.

Mehta

. Pioneering the future of mental health with AI: exploring promising directions. In: Papadopoulou

Lytras

Konstantinopoulou

(eds) Advances in medical technologies and clinical practice. USA: IGI Global, 2024, pp.277–304. doi:10.4018/979-8-3693-4422-4.ch014.

36.

Porritt

Gomersall

Lockwood

. JBI’s systematic reviews: study selection and critical appraisal. AJN, American Journal of Nursing 2014; 114: 47–52.

37.

Banos

Comas-González

Medina

, et al. Sensing technologies and machine learning methods for emotion recognition in autism: systematic review. Int J Med Inf 2024; 187: 105469.

38.

Rebelo

Verboom

Dos Santos

, et al. The impact of artificial intelligence on the tasks of mental healthcare workers: a scoping review. Computers in Human Behavior: Artificial Humans 2023; 1: 100008.

39.

Thenral

Annamalai

. Telepsychiatry and the role of artificial intelligence in mental health in post-COVID-19 India: a scoping review on opportunities. Indian J Psychol Med 2020; 42: 428–434.

40.

Bouchouras

Kotis

. Integrating artificial intelligence, internet of things, and sensor-based technologies: a systematic review of methodologies in autism Spectrum disorder detection. Algorithms 2025; 18: 34.

41.

Cruz-Gonzalez

AWJ

Lam

, et al. Artificial intelligence in mental health care: a systematic review of diagnosis, monitoring, and intervention applications. Psychol Med 2025; 55: 18.

42.

Tay

Htun

Sim

. Prediction of clinical outcomes in psychotic disorders using artificial intelligence methods: a scoping review. Brain Sci 2024; 14: 78.

43.

De Belen

RAJ

Bednarz

Sowmya

, et al. Computer vision in autism spectrum disorder research: a systematic review of published studies from 2009 to 2019. Transl Psychiatry 2020; 10: 33.

44.

Higgins

Short

Chalup

, et al. Artificial intelligence (AI) and machine learning (ML) based decision support systems in mental health: an integrative review. Int J Mental Health Nurs 2023; 32: 966–978.

45.

Milasan

Scott-Purdy

. The future of artificial intelligence in mental health nursing practice: an integrative review. Int J Mental Health Nurs 2025; 34: e70003.

46.

Alqaysi

Albahri

Hamid

. Diagnosis-Based hybridization of multimedical tests and sociodemographic characteristics of autism Spectrum disorder using artificial intelligence and machine learning techniques: a systematic review. Int J Telemed Appl 2022; 2022: 1–26. Hu F, ed.

47.

Scoglio

Reilly

Gorman

, et al. Use of social robots in mental health and well-being research: systematic review. J Med Internet Res 2019; 21: e13322.

48.

Le Glaz

Haralambous

Kim-Dufor

, et al. Machine learning and natural language processing in mental health: systematic review. J Med Internet Res 2021; 23: e15708.

49.

Abd-alrazaq

AlSaad

Aziz

, et al. Wearable artificial intelligence for anxiety and depression: scoping review. J Med Internet Res 2023; 25: e42672.

50.

Auf

Svedberg

Nygren

, et al. The use of AI in mental health services to support decision-making: scoping review. J Med Internet Res 2025; 27: e63548.

51.

Milne-Ives

Selby

Inkster

, et al. Artificial intelligence and machine learning in mobile apps for mental health: a scoping review. PLOS Digit Health 2022; 1: e0000079. Narasimhan P, ed.

52.

Cavus

Lawan

Ibrahim

, et al. A systematic literature review on the application of machine-learning models in behavioral assessment of autism Spectrum disorder. JPM 2021; 11: 99.

53.

Moon

Hwang

Kana

, et al. Accuracy of machine learning algorithms for the diagnosis of autism Spectrum disorder: systematic review and meta-analysis of brain magnetic resonance imaging studies. JMIR Ment Health 2019; 6: e14108.

54.

Rogan

Bucci

Firth

. Health care Professionals’ views on the use of passive sensing, AI, and machine learning in mental health care: systematic review with meta-synthesis. JMIR Ment Health 2024; 11: e49577.

55.

Rahsepar Meadi

Sillekens

Metselaar

, et al. Exploring the ethical challenges of conversational AI in mental health care: scoping review. JMIR Ment Health 2025; 12: e60432.

56.

Scherbakov

Hubig

Lenert

, et al. Natural language processing and social determinants of health in mental health research: aI-assisted scoping review. JMIR Ment Health 2025; 12: e67192–e67192.

57.

Cecula

Dawoodbhoy

, et al. Applications of artificial intelligence to improve patient flow on mental health inpatient units - narrative literature review. Heliyon 2021; 7: e06626.

58.

Bhatt

Liu

Gong

, et al. Emerging artificial intelligence–empowered mHealth: scoping review. JMIR Mhealth Uhealth 2022; 10: e35053.

59.

Ding

Zhang

Qiu

. Deep learning approach to predict autism spectrum disorder: a systematic review and meta-analysis. BMC Psychiatry 2024; 24: 39.

60.

Pathak

, et al. Deep learning in mental health outcome research: a scoping review. Transl Psychiatry 2020; 10: 16.

61.

Islam

Shahriyar

Agarwala

, et al. Artificial intelligence-based risk assessment tools for sexual, reproductive and mental health: a systematic review. BMC Med Inform Decis Mak 2025; 25: 32.

62.

Alabdulkareem

Alhakbani

Al-Nafjan

. A systematic review of research on robot-assisted therapy for children with autism. Sensors 2022; 22: 44.

63.

Solek

Nurfitri

Sahril

, et al. The role of artificial intelligence for early diagnostic tools of autism Spectrum disorder: a systematic review. Turkish Archives of Pediatrics 2025; 60: 126–140.

64.

Kohli

Kar

Sinha

. The role of intelligent technologies in early detection of autism Spectrum disorder (ASD): a scoping review. IEEE Access 2022; 10: 104887–104913.

65.

Kolding

Lundin

Hansen

, et al. Use of generative artificial intelligence (AI) in psychiatry and mental health care: a systematic review. Acta Neuropsychiatr 2025; 37: 37.

66.

Zhang

Lee

, et al. Systematic review and meta-analysis of AI-based conversational agents for promoting mental health and well-being. npj Digit Med 2023; 6: 36.

67.

Page

McKenzie

Bossuyt

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. Br Med J 2021; 372: n71. Published online March 29, 2021. doi:10.1136/bmj.n71

68.

Liu

Sun

, et al. Bridging the gap between artificial intelligence and mental health. Science Bulletin 2023; 68: 1606–1610.

69.

Wall

Dally

Luyster

, et al. Use of artificial intelligence to shorten the behavioral diagnosis of autism. PLoS ONE 2012; 7: e43855. Rogers S, ed.

70.

Montag

Ali

Al-Thani

, et al. On artificial intelligence and global mental health. Asian J Psychiatr 2024; 91: 103855.

71.

Al Banna

Ghosh

Taher

, et al. A monitoring system for patients of autism Spectrum disorder using artificial intelligence. In: Mahmud

Vassanelli

Kaiser

, etal (eds) Brain informatics. Lecture notes in computer science. 12241. Switzerland: Springer International Publishing, 2020, pp.251–262. doi:10.1007/978-3-030-59277-6_23.

72.

Song

Kim

Bong

, et al. The use of artificial intelligence in screening and diagnosis of autism Spectrum disorder: a literature review. J Korean Acad Child Adolesc Psychiatry 2019; 30: 145–152.

73.

Shahamiri

Thabtah

Abdelhamid

. A new classification system for autism based on machine learning of artificial intelligence. THC 2022; 30: 605–622.

74.

Masri

Mat Jani

. Employing artificial intelligence techniques in mental health diagnostic expert system2012 International conference on computer & information science (ICCIS). Malaysia: IEEE, 2012, pp.495–499. doi:10.1109/ICCISci.2012.6297296.

75.

Rubeis

. Ihealth: the ethics of artificial intelligence and big data in mental healthcare. Internet Interv 2022; 28: 100518.

76.

Liang

Zheng

, et al. Exploration and strategy analysis of mental health education for students in sports majors in the era of artificial intelligence. Front Psychol 2022; 12: 762725.

77.

Rosenfeld

Benrimoh

Armstrong

, et al.

Big data analytics and artificial intelligence in mental healthcare

Applications of big data in healthcare. Netherlands: Elsevier, 2021, pp.137–171. doi:10.1016/B978-0-12-820203-6.00001-1.

78.

Danieli

Ciulli

Mousavi

, et al. A conversational artificial intelligence agent for a mental health care app: evaluation study of its participatory design. JMIR Form Res 2021; 5: e30053.

79.

Kleine

Kokje

Lermer

, et al. Attitudes toward the adoption of 2 artificial intelligence–enabled mental health tools among prospective psychotherapists: cross-sectional study. JMIR Hum Factors 2023; 10: e46859.

80.

Ćosić

Popović

Šarlija

, et al. Artificial intelligence in prediction of mental health disorders induced by the COVID-19 pandemic among health care workers. Croat Med J 2020; 61: 279–288.

81.

Golden

Popescu

Israel

, et al.

Applying artificial intelligence to clinical decision support in mental health: what have we learned?

Health Policy Technol 2024; 13: 100844.

82.

Shan

Xie

, et al. Public trust in artificial intelligence applications in mental health care: topic modeling analysis. JMIR Hum Factors 2022; 9: e38799.

83.

Cabrera

Loyola

Magaña

, et al. Ethical dilemmas, mental health, artificial intelligence, and LLM-based chatbots. In: Rojas

Valenzuela

Rojas Ruiz

, etal (eds) Bioinformatics and biomedical engineering. Lecture notes in computer science. 13920. Switzerland: Springer Nature Switzerland, 2023, pp.313–326. doi:10.1007/978-3-031-34960-7_22.

84.

Van Schalkwyk

. Artificial intelligence in pediatric behavioral health. Child Adolesc Psychiatry Ment Health 2023; 17: 38. s13034-023-00586-y.

85.

Sinha

Meheli

Kadaba

. Understanding digital mental health needs and usage with an artificial intelligence–led mental health app (wysa) during the COVID-19 pandemic: retrospective analysis. JMIR Form Res 2023; 7: e41913.

86.

Hilty

Cheng

Luxton

. Artificial intelligence and predictive modeling in mental health. In: Mucić

Hilty

(eds) Digital mental health. Switzerland: Springer Nature Switzerland, 2024, pp.323–350. doi:10.1007/978-3-031-59936-1_13.

87.

Chen

Yuan

Dong

, et al. Artificial intelligence significantly facilitates development in the mental health of college students: a bibliometric analysis. Front Psychol 2024; 15: 1375294.

88.

Nilsen

Svedberg

Nygren

, et al. Accelerating the impact of artificial intelligence in mental healthcare through implementation science. Implementation Research and Practice 2022; 3: 26334895221112033.

89.

Wang

Chen

, et al. Integration of artificial intelligence and wearable internet of things for mental health detection. International Journal of Cognitive Computing in Engineering 2024; 5: 307–315.

90.

Wilson

Higgins

Atem

, et al. Artificial intelligence: an eye cast towards the mental health nursing horizon. Int J Mental Health Nurs 2023; 32: 938–944.

91.

Saeidnia

Hashemi Fotami

Lund

, et al. Ethical considerations in artificial intelligence interventions for mental health and well-being: ensuring responsible implementation and impact. Soc Sci 2024; 13: 81.

92.

Zidaru

Morrow

Stockley

. Ensuring patient and public involvement in the transition to AI-assisted mental health care: a systematic scoping review and agenda for design justice. Health Expect 2021; 24: 1072–1124.

93.

Gooding

Kariotis

. Ethics and law in research on algorithmic and data-driven technology in mental health care: scoping review. JMIR Ment Health 2021; 8: e24668.

94.

Delanerolle

Yang

Shetty

, et al. Artificial intelligence: a rapid case for advancement in the personalization of gynaecology/obstetric and mental health care. Womens Health (Lond Engl) 2021; 17: 17455065211018111.

95.

Jain

Patil

Dutt

, et al.

Contribution of artificial intelligence to the promotion of mental health

2022 5th international conference on contemporary computing and informatics (IC3I). India: IEEE, 2022, pp.1938–1944. doi:10.1109/IC3I56241.2022.10073302.

96.

Drydakis

. Mobile applications aiming to facilitate immigrants’ societal integration and overall level of integration, health and mental health. Does artificial intelligence enhance outcomes? Comput Human Behav 2021; 117: 106661.

97.

Andrew

Rudra

Eunice

, et al. Artificial intelligence in adolescents mental health disorder diagnosis, prognosis, and treatment. Front Public Health 2023; 11: 1110088.

98.

Connors

Janse

De Jong

, et al. The use of feedback in mental health services: expanding horizons on reach and implementation. Adm Policy Ment Health 2025; 52: 1–10.

99.

Graham

Depp

. Artificial intelligence and risk prediction in geriatric mental health: what happens next? Int Psychogeriatr 2019; 31: 921–923.

100.

Gamble

. Artificial intelligence and mobile apps for mental healthcare: a social informatics perspective. AJIM 2020; 72: 509–523.

101.

Van Heerden

Pozuelo

Kohrt

. Global mental health services and the impact of artificial intelligence–powered large language models. JAMA Psychiatry 2023; 80: 62.

102.

Götzl

Hiller

Rauschenberg

, et al. Artificial intelligence-informed mobile mental health apps for young people: a mixed-methods approach on users’ and stakeholders’ perspectives. Child Adolesc Psychiatry Ment Health 2022; 16: 86.

103.

Alibudbud

Aruta

JJBR

Sison

, et al. Artificial intelligence in the era of planetary health: insights on its application for the climate change-mental health nexus in the Philippines. Int Rev Psychiatry 2025; 37: 21–32.

104.

Warrier

Khandelwal

. Ethical considerations in the use of artificial intelligence in mental health. Egypt J Neurol Psychiatry Neurosurg 2023; 59: 39.

105.

Solaiman

Malik

Ghuloum

. Monitoring mental health: legal and ethical considerations of using artificial intelligence in psychiatric wards. Am J Law Med 2023; 49: 250–266.

The role of AI-driven art therapy in supporting autism,mental health,and emotional well-being: An umbrella review

Abstract

Background

Aim

Methods

Results

Conclusions

Keywords

Introduction

Methodology

Research design

Literature search strategy

Inclusion and exclusion criteria

Screening and selection procedure

Quality assessment

Data extraction process

Thematic analysis and data synthesis

Reporting standards

Results

Overview of included reviews

Effectiveness and therapeutic benefits of AI-driven art therapy

Specific impacts on autism spectrum disorders

Improvements in mental health

Enhancements in emotional well-being

Technological considerations and implementation barriers

User experiences and accessibility factors

Discussion

Implications of the study

Limitations

Recommendations for future research

Conclusion

Footnotes

Acknowledgments

ORCID iD

Ethical approval

Contributorship

Funding

Declaration of conflict of interest

References