Jmir Mental Health最新文献

{"title":"Responsible Design, Integration, and Use of Generative AI in Mental Health.","authors":"Oren Asman, John Torous, Amir Tal","doi":"10.2196/70439","DOIUrl":"10.2196/70439","url":null,"abstract":"<p><strong>Unlabelled: </strong>Generative artificial intelligence (GenAI) shows potential for personalized care, psychoeducation, and even crisis prediction in mental health, yet responsible use requires ethical consideration and deliberation and perhaps even governance. This is the first published theme issue focused on responsible GenAI in mental health. It brings together evidence and insights on GenAI's capabilities, such as emotion recognition, therapy-session summarization, and risk assessment, while highlighting the sensitive nature of mental health data and the need for rigorous validation. Contributors discuss how bias, alignment with human values, transparency, and empathy must be carefully addressed to ensure ethically grounded, artificial intelligence-assisted care. By proposing conceptual frameworks; best practices; and regulatory approaches, including ethics of care and the preservation of socially important humanistic elements, this theme issue underscores that GenAI can complement, rather than replace, the vital role of human empathy in clinical settings. To achieve this, an ongoing collaboration between researchers, clinicians, policy makers, and technologists is essential.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e70439"},"PeriodicalIF":4.8,"publicationDate":"2025-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11769776/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143048089","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Natural Language Processing and Social Determinants of Health in Mental Health Research: AI-Assisted Scoping Review.","authors":"Dmitry A Scherbakov, Nina C Hubig, Leslie A Lenert, Alexander V Alekseyenko, Jihad S Obeid","doi":"10.2196/67192","DOIUrl":"10.2196/67192","url":null,"abstract":"<p><strong>Background: </strong>The use of natural language processing (NLP) in mental health research is increasing, with a wide range of applications and datasets being investigated.</p><p><strong>Objective: </strong>This review aims to summarize the use of NLP in mental health research, with a special focus on the types of text datasets and the use of social determinants of health (SDOH) in NLP projects related to mental health.</p><p><strong>Methods: </strong>The search was conducted in September 2024 using a broad search strategy in PubMed, Scopus, and CINAHL Complete. All citations were uploaded to Covidence (Veritas Health Innovation) software. The screening and extraction process took place in Covidence with the help of a custom large language model (LLM) module developed by our team. This LLM module was calibrated and tuned to automate many aspects of the review process.</p><p><strong>Results: </strong>The screening process, assisted by the custom LLM, led to the inclusion of 1768 studies in the final review. Most of the reviewed studies (n=665, 42.8%) used clinical data as their primary text dataset, followed by social media datasets (n=523, 33.7%). The United States contributed the highest number of studies (n=568, 36.6%), with depression (n=438, 28.2%) and suicide (n=240, 15.5%) being the most frequently investigated mental health issues. Traditional demographic variables, such as age (n=877, 56.5%) and gender (n=760, 49%), were commonly extracted, while SDOH factors were less frequently reported, with urban or rural status being the most used (n=19, 1.2%). Over half of the citations (n=826, 53.2%) did not provide clear information on dataset accessibility, although a sizable number of studies (n=304, 19.6%) made their datasets publicly available.</p><p><strong>Conclusions: </strong>This scoping review underscores the significant role of clinical notes and social media in NLP-based mental health research. Despite the clear relevance of SDOH to mental health, their underutilization presents a gap in current research. This review can be a starting point for researchers looking for an overview of mental health projects using text data. Shared datasets could be used to place more emphasis on SDOH in future studies.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e67192"},"PeriodicalIF":4.8,"publicationDate":"2025-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11756842/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143014583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Determinants of Dropout From a Virtual Agent-Based App for Insomnia Management in a Self-Selected Sample of Users With Insomnia Symptoms: Longitudinal Study.","authors":"María Montserrat Sanchez Ortuño, Florian Pecune, Julien Coelho, Jean Arthur Micoulaud-Franchi, Nathalie Salles, Marc Auriacombe, Fuschia Serre, Yannick Levavasseur, Etienne De Sevin, Patricia Sagaspe, Pierre Philip","doi":"10.2196/51022","DOIUrl":"10.2196/51022","url":null,"abstract":"<p><strong>Background: </strong>Fully automated digital interventions delivered via smartphone apps have proven efficacious for a wide variety of mental health outcomes. An important aspect is that they are accessible at a low cost, thereby increasing their potential public impact and reducing disparities. However, a major challenge to their successful implementation is the phenomenon of users dropping out early.</p><p><strong>Objective: </strong>The purpose of this study was to pinpoint the factors influencing early dropout in a sample of self-selected users of a virtual agent (VA)-based behavioral intervention for managing insomnia, named KANOPEE, which is freely available in France.</p><p><strong>Methods: </strong>From January 2021 to December 2022, of the 9657 individuals, aged 18 years or older, who downloaded and completed the KANOPEE screening interview and had either subclinical or clinical insomnia symptoms, 4295 (44.5%) dropped out (ie, did not return to the app to continue filling in subsequent assessments). The primary outcome was a binary variable: having dropped out after completing the screening assessment (early dropout) or having completed all the treatment phases (n=551). Multivariable logistic regression analysis was used to identify predictors of dropout among a set of sociodemographic, clinical, and sleep diary variables, and users' perceptions of the treatment program, collected during the screening interview.</p><p><strong>Results: </strong>The users' mean age was 47.95 (SD 15.21) years. Of those who dropped out early and those who completed the treatment, 65.1% (3153/4846) were women and 34.9% (1693/4846) were men. Younger age (adjusted odds ratio [AOR] 0.98, 95% CI 0.97-0.99), lower education level (compared to middle school; high school: AOR 0.56, 95% CI 0.35-0.90; bachelor's degree: AOR 0.35, 95% CI 0.23-0.52; master's degree or higher: AOR 0.35, 95% CI 0.22-0.55), poorer nocturnal sleep (sleep efficiency: AOR 0.64, 95% CI 0.42-0.96; number of nocturnal awakenings: AOR 1.13, 95% CI 1.04-1.23), and more severe depression symptoms (AOR 1.12, 95% CI 1.04-1.21) were significant predictors of dropping out. When measures of perceptions of the app were included in the model, perceived benevolence and credibility of the VA decreased the odds of dropout (AOR 0.91, 95% CI 0.85-0.97).</p><p><strong>Conclusions: </strong>As in traditional face-to-face cognitive behavioral therapy for insomnia, the presence of significant depression symptoms plays an important role in treatment dropout. This variable represents an important target to address to increase early engagement with fully automated insomnia management programs. Furthermore, our results support the contention that a VA can provide relevant user stimulation that will eventually pay out in terms of user engagement.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e51022"},"PeriodicalIF":4.8,"publicationDate":"2025-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11753580/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143014536","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Digital integrated interventions for comorbid depression and substance use disorder: narrative review and content analysis.","authors":"Geneva K Jonathan, Qiuzuo Guo, Heyli Arcese, A Eden Evins, Sabine Wilhelm","doi":"10.2196/67670","DOIUrl":"10.2196/67670","url":null,"abstract":"<p><strong>Background: </strong>Integrated digital interventions for the treatment of comorbid depression and substance use disorder have been developed, and evidence of their effectiveness is mixed.</p><p><strong>Objective: </strong>To better understand the potential underlying causes of these mixed findings, we described intervention characteristics, examined evidence-based treatment strategies within integrated digital treatments, reported the frequency of specific evidence-based strategies across different treatment modalities, and identified overlap between various treatment strategies and critical gaps in existing literature.</p><p><strong>Methods: </strong>In June 2024, a literature search was conducted in Google Scholar to identify digital integrated interventions for comorbid MDD and SUD. Articles were included if they described interventions targeting both conditions simultaneously, were grounded in CBT, MI, or MET, and were delivered at least in part via digital modalities. Fourteen studies meeting these criteria were coded using an open coding approach to identify treatment strategies. Statistical analyses summarized the number, frequency, and overlap of these strategies.</p><p><strong>Results: </strong>Half of studies (50.0%, n=7) included participants with mild to moderate depression symptom severity and hazardous substance use. Only 35.7% (n=5) of the studies required that participants meet the full diagnostic criteria for MDD, as assessed by the SCID or MINI, for inclusion and 21.4% (n=3) required a SUD diagnosis. Web-based (35.3%, n=6), computer-based (21.4%, n=3) and supportive text messaging interventions (21.4%, n=3) were included. Treatment duration averaged 10.3 weeks (SD=6.8). Common treatment strategies included self-monitoring (78.6%, n=11), psychoeducation (71.4%, n=10), and coping skills (64.3%, n=9). Interventions often combined therapeutic strategies, with psychoeducation frequently paired with self-monitoring (64.3%, n=9) and coping skills (50%, n=7).</p><p><strong>Conclusions: </strong>Among integrated digital interventions for comorbid depression and substance use, there was significant variability in inclusion criteria, digital modalities, methodology, and treatment strategies, significant methodological challenges, and underrepresentation of evidence-based practices. Without standardized methodologies comparison of the clinical outcomes across studies is challenging. These results emphasize the critical need for future research to adopt standardized approaches, thereby facilitating more accurate comparisons and a deeper understanding of intervention efficacy.</p><p><strong>Clinicaltrial: </strong></p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":" ","pages":""},"PeriodicalIF":4.8,"publicationDate":"2025-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143651471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Utility of Consumer-Grade Wearable Devices for Inferring Physical and Mental Health Outcomes in Severe Mental Illness: Systematic Review.","authors":"Lamiece Hassan, Alyssa Milton, Chelsea Sawyer, Alexander J Casson, John Torous, Alan Davies, Bernalyn Ruiz-Yu, Joseph Firth","doi":"10.2196/65143","DOIUrl":"10.2196/65143","url":null,"abstract":"<p><strong>Background: </strong>Digital wearable devices, worn on or close to the body, have potential for passively detecting mental and physical health symptoms among people with severe mental illness (SMI); however, the roles of consumer-grade devices are not well understood.</p><p><strong>Objective: </strong>This study aims to examine the utility of data from consumer-grade, digital, wearable devices (including smartphones or wrist-worn devices) for remotely monitoring or predicting changes in mental or physical health among adults with schizophrenia or bipolar disorder. Studies were included that passively collected physiological data (including sleep duration, heart rate, sleep and wake patterns, or physical activity) for at least 3 days. Research-grade actigraphy methods and physically obtrusive devices were excluded.</p><p><strong>Methods: </strong>We conducted a systematic review of the following databases: Cochrane Central Register of Controlled Trials, Technology Assessment, AMED (Allied and Complementary Medicine), APA PsycINFO, Embase, MEDLINE(R), and IEEE XPlore. Searches were completed in May 2024. Results were synthesized narratively due to study heterogeneity and divided into the following phenotypes: physical activity, sleep and circadian rhythm, and heart rate.</p><p><strong>Results: </strong>Overall, 23 studies were included that reported data from 12 distinct studies, mostly using smartphones and centered on relapse prevention. Only 1 study explicitly aimed to address physical health outcomes among people with SMI. In total, data were included from over 500 participants with SMI, predominantly from high-income countries. Most commonly, papers presented physical activity data (n=18), followed by sleep and circadian rhythm data (n=14) and heart rate data (n=6). The use of smartwatches to support data collection were reported by 8 papers; the rest used only smartphones. There was some evidence that lower levels of activity, higher heart rates, and later and irregular sleep onset times were associated with psychiatric diagnoses or poorer symptoms. However, heterogeneity in devices, measures, sampling and statistical approaches complicated interpretation.</p><p><strong>Conclusions: </strong>Consumer-grade wearables show the ability to passively detect digital markers indicative of psychiatric symptoms or mental health status among people with SMI, but few are currently using these to address physical health inequalities. The digital phenotyping field in psychiatry would benefit from moving toward agreed standards regarding data descriptions and outcome measures and ensuring that valuable temporal data provided by wearables are fully exploited.</p><p><strong>Trial registration: </strong>PROSPERO CRD42022382267; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=382267.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e65143"},"PeriodicalIF":4.8,"publicationDate":"2025-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11751658/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142957107","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"User Experiences of and Preferences for Self-Guided Digital Interventions for the Treatment of Mild to Moderate Eating Disorders: Systematic Review and Metasynthesis.","authors":"Lauryn Gar-Mun Cheung, Pamela Carien Thomas, Eva Brvar, Sarah Rowe","doi":"10.2196/57795","DOIUrl":"10.2196/57795","url":null,"abstract":"<p><strong>Background: </strong>Digital interventions typically involve using smartphones or PCs to access online or downloadable self-help and may offer a more accessible and convenient option than face-to-face interventions for some people with mild to moderate eating disorders. They have been shown to substantially reduce eating disorder symptoms, but treatment dropout rates are higher than for face-to-face interventions. We need to understand user experiences and preferences for digital interventions to support the design and development of user-centered digital interventions that are engaging and meet users' needs.</p><p><strong>Objective: </strong>This study aims to understand user experiences and user preferences for digital interventions that aim to reduce mild to moderate eating disorder symptoms in adults.</p><p><strong>Methods: </strong>We conducted a metasynthesis of qualitative studies. We searched 6 databases for published and unpublished literature from 2013 to 2024. We searched for studies conducted in naturalistic or outpatient settings, using primarily unguided digital self-help interventions designed to reduce eating disorder symptoms in adults with mild to moderate eating disorders. We conducted a thematic synthesis using line-by-line coding of the results and findings from each study to generate themes.</p><p><strong>Results: </strong>A total of 8 studies were included after screening 3695 search results. Overall, 7 metathemes were identified. The identified metathemes included the appeal of digital interventions, role of digital interventions in treatment, value of support in treatment, communication at the right level, importance of engagement, shaping knowledge to improve eating disorder behaviors, and design of the digital intervention. Users had positive experiences with digital interventions and perceived them as helpful for self-reflection and mindfulness. Users found digital interventions to be convenient and flexible and that they fit with their lifestyle. Overall, users noticed reduced eating disorder thoughts and behaviors. However, digital interventions were not generally perceived as a sufficient treatment that could replace traditional face-to-face treatment. Users have individual needs, so an ideal intervention would offer personalized content and functions.</p><p><strong>Conclusions: </strong>Users found digital interventions for eating disorders practical and effective but stressed the need for interventions to address the full range of symptoms, severity, and individual needs. Future digital interventions should be cocreated with users and offer more personalization. Further research is needed to determine the appropriate balance of professional and peer support and whether these interventions should serve as the first step in the stepped care model.</p><p><strong>Trial registration: </strong>PROSPERO CRD42023426932; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=426932.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"12 ","pages":"e57795"},"PeriodicalIF":4.8,"publicationDate":"2025-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11748441/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142922080","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing Health Technology Literacy and Attitudes of Patients in an Urban Outpatient Psychiatry Clinic: Cross-Sectional Survey Study.","authors":"Julia Tartaglia, Brendan Jaghab, Mohamed Ismail, Katrin Hänsel, Anna Van Meter, Michael Kirschenbaum, Michael Sobolev, John M Kane, Sunny X Tang","doi":"10.2196/63034","DOIUrl":"10.2196/63034","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Digital health technologies are increasingly being integrated into mental health care. However, the adoption of these technologies can be influenced by patients' digital literacy and attitudes, which may vary based on sociodemographic factors. This variability necessitates a better understanding of patient digital literacy and attitudes to prevent a digital divide, which can worsen existing health care disparities.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This study aimed to assess digital literacy and attitudes toward digital health technologies among a diverse psychiatric outpatient population. In addition, the study sought to identify clusters of patients based on their digital literacy and attitudes, and to compare sociodemographic characteristics among these clusters.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;A survey was distributed to adult psychiatric patients with various diagnoses in an urban outpatient psychiatry program. The survey included a demographic questionnaire, a digital literacy questionnaire, and a digital health attitudes questionnaire. Multiple linear regression analyses were used to identify predictors of digital literacy and attitudes. Cluster analysis was performed to categorize patients based on their responses. Pairwise comparisons and one-way ANOVA were conducted to analyze differences between clusters.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;A total of 256 patients were included in the analysis. The mean age of participants was 32 (SD 12.6, range 16-70) years. The sample was racially and ethnically diverse: White (100/256, 38.9%), Black (39/256, 15.2%), Latinx (44/256, 17.2%), Asian (59/256, 23%), and other races and ethnicities (15/256, 5.7%). Digital literacy was high for technologies such as smartphones, videoconferencing, and social media (items with &gt;75%, 193/256 of participants reporting at least some use) but lower for health apps, mental health apps, wearables, and virtual reality (items with &lt;42%, 108/256 reporting at least some use). Attitudes toward using technology in clinical care were generally positive (9 out of 10 items received &gt;75% positive score), particularly for communication with providers and health data sharing. Older age (P&lt;.001) and lower educational attainment (P&lt;.001) negatively predicted digital literacy scores, but no demographic variables predicted attitude scores. Cluster analysis identified 3 patient groups. Relative to the other clusters, cluster 1 (n=30) had lower digital literacy and intermediate acceptance of digital technology. Cluster 2 (n=50) had higher literacy and lower acceptance. Cluster 3 (n=176) displayed both higher literacy and acceptance. Significant between-cluster differences were observed in mean age and education level between clusters (P&lt;.001), with cluster 1 participants being older and having lower levels of formal education.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;High digital literacy and acceptance of digital technologies were observed among our patients,","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"11 ","pages":"e63034"},"PeriodicalIF":4.8,"publicationDate":"2024-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11729776/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142928393","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Momentary Depression Severity Prediction in Patients With Acute Depression Who Undergo Sleep Deprivation Therapy: Speech-Based Machine Learning Approach.","authors":"Lisa-Marie Hartnagel, Daniel Emden, Jerome C Foo, Fabian Streit, Stephanie H Witt, Josef Frank, Matthias F Limberger, Sara E Schmitz, Maria Gilles, Marcella Rietschel, Tim Hahn, Ulrich W Ebner-Priemer, Lea Sirignano","doi":"10.2196/64578","DOIUrl":"10.2196/64578","url":null,"abstract":"<p><strong>Background: </strong>Mobile devices for remote monitoring are inevitable tools to support treatment and patient care, especially in recurrent diseases such as major depressive disorder. The aim of this study was to learn if machine learning (ML) models based on longitudinal speech data are helpful in predicting momentary depression severity. Data analyses were based on a dataset including 30 inpatients during an acute depressive episode receiving sleep deprivation therapy in stationary care, an intervention inducing a rapid change in depressive symptoms in a relatively short period of time. Using an ambulatory assessment approach, we captured speech samples and assessed concomitant depression severity via self-report questionnaire over the course of 3 weeks (before, during, and after therapy). We extracted 89 speech features from the speech samples using the Extended Geneva Minimalistic Acoustic Parameter Set from the Open-Source Speech and Music Interpretation by Large-Space Extraction (audEERING) toolkit and the additional parameter speech rate.</p><p><strong>Objective: </strong>We aimed to understand if a multiparameter ML approach would significantly improve the prediction compared to previous statistical analyses, and, in addition, which mechanism for splitting training and test data was most successful, especially focusing on the idea of personalized prediction.</p><p><strong>Methods: </strong>To do so, we trained and evaluated a set of >500 ML pipelines including random forest, linear regression, support vector regression, and Extreme Gradient Boosting regression models and tested them on 5 different train-test split scenarios: a group 5-fold nested cross-validation at the subject level, a leave-one-subject-out approach, a chronological split, an odd-even split, and a random split.</p><p><strong>Results: </strong>In the 5-fold cross-validation, the leave-one-subject-out, and the chronological split approaches, none of the models were statistically different from random chance. The other two approaches produced significant results for at least one of the models tested, with similar performance. In total, the superior model was an Extreme Gradient Boosting in the odd-even split approach (R²=0.339, mean absolute error=0.38; both P<.001), indicating that 33.9% of the variance in depression severity could be predicted by the speech features.</p><p><strong>Conclusions: </strong>Overall, our analyses highlight that ML fails to predict depression scores of unseen patients, but prediction performance increased strongly compared to our previous analyses with multilevel models. We conclude that future personalized ML models might improve prediction performance even more, leading to better patient management and care.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"11 ","pages":"e64578"},"PeriodicalIF":4.8,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11684135/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142878227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Balancing Between Privacy and Utility for Affect Recognition Using Multitask Learning in Differential Privacy-Added Federated Learning Settings: Quantitative Study.","authors":"Mohamed Benouis, Elisabeth Andre, Yekta Said Can","doi":"10.2196/60003","DOIUrl":"10.2196/60003","url":null,"abstract":"<p><strong>Background: </strong>The rise of wearable sensors marks a significant development in the era of affective computing. Their popularity is continuously increasing, and they have the potential to improve our understanding of human stress. A fundamental aspect within this domain is the ability to recognize perceived stress through these unobtrusive devices.</p><p><strong>Objective: </strong>This study aims to enhance the performance of emotion recognition using multitask learning (MTL), a technique extensively explored across various machine learning tasks, including affective computing. By leveraging the shared information among related tasks, we seek to augment the accuracy of emotion recognition while confronting the privacy threats inherent in the physiological data captured by these sensors.</p><p><strong>Methods: </strong>To address the privacy concerns associated with the sensitive data collected by wearable sensors, we proposed a novel framework that integrates differential privacy and federated learning approaches with MTL. This framework was designed to efficiently identify mental stress while preserving private identity information. Through this approach, we aimed to enhance the performance of emotion recognition tasks while preserving user privacy.</p><p><strong>Results: </strong>Comprehensive evaluations of our framework were conducted using 2 prominent public datasets. The results demonstrate a significant improvement in emotion recognition accuracy, achieving a rate of 90%. Furthermore, our approach effectively mitigates privacy risks, as evidenced by limiting reidentification accuracies to 47%.</p><p><strong>Conclusions: </strong>This study presents a promising approach to advancing emotion recognition capabilities while addressing privacy concerns in the context of empathetic sensors. By integrating MTL with differential privacy and federated learning, we have demonstrated the potential to achieve high levels of accuracy in emotion recognition while ensuring the protection of user privacy. This research contributes to the ongoing efforts to use affective computing in a privacy-aware and ethical manner.</p>","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"11 ","pages":"e60003"},"PeriodicalIF":4.8,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11684349/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142878300","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Early Attrition Prediction for Web-Based Interpretation Bias Modification to Reduce Anxious Thinking: A Machine Learning Study.","authors":"Sonia Baee, Jeremy W Eberle, Anna N Baglione, Tyler Spears, Elijah Lewis, Hongning Wang, Daniel H Funk, Bethany Teachman, Laura E Barnes","doi":"10.2196/51567","DOIUrl":"10.2196/51567","url":null,"abstract":"&lt;p&gt;&lt;strong&gt;Background: &lt;/strong&gt;Digital mental health is a promising paradigm for individualized, patient-driven health care. For example, cognitive bias modification programs that target interpretation biases (cognitive bias modification for interpretation [CBM-I]) can provide practice thinking about ambiguous situations in less threatening ways on the web without requiring a therapist. However, digital mental health interventions, including CBM-I, are often plagued with lack of sustained engagement and high attrition rates. New attrition detection and mitigation strategies are needed to improve these interventions.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Objective: &lt;/strong&gt;This paper aims to identify participants at a high risk of dropout during the early stages of 3 web-based trials of multisession CBM-I and to investigate which self-reported and passively detected feature sets computed from the participants interacting with the intervention and assessments were most informative in making this prediction.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Methods: &lt;/strong&gt;The participants analyzed in this paper were community adults with traits such as anxiety or negative thinking about the future (Study 1: n=252, Study 2: n=326, Study 3: n=699) who had been assigned to CBM-I conditions in 3 efficacy-effectiveness trials on our team's public research website. To identify participants at a high risk of dropout, we created 4 unique feature sets: self-reported baseline user characteristics (eg, demographics), self-reported user context and reactions to the program (eg, state affect), self-reported user clinical functioning (eg, mental health symptoms), and passively detected user behavior on the website (eg, time spent on a web page of CBM-I training exercises, time of day during which the exercises were completed, latency of completing the assessments, and type of device used). Then, we investigated the feature sets as potential predictors of which participants were at high risk of not starting the second training session of a given program using well-known machine learning algorithms.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Results: &lt;/strong&gt;The extreme gradient boosting algorithm performed the best and identified participants at high risk with macro-F&lt;sub&gt;1&lt;/sub&gt;-scores of .832 (Study 1 with 146 features), .770 (Study 2 with 87 features), and .917 (Study 3 with 127 features). Features involving passive detection of user behavior contributed the most to the prediction relative to other features. The mean Gini importance scores for the passive features were as follows: .033 (95% CI .019-.047) in Study 1; .029 (95% CI .023-.035) in Study 2; and .045 (95% CI .039-.051) in Study 3. However, using all features extracted from a given study led to the best predictive performance.&lt;/p&gt;&lt;p&gt;&lt;strong&gt;Conclusions: &lt;/strong&gt;These results suggest that using passive indicators of user behavior, alongside self-reported measures, can improve the accuracy of prediction of participants at a high risk of dropout early during multisession CBM-I programs","PeriodicalId":48616,"journal":{"name":"Jmir Mental Health","volume":"11 ","pages":"e51567"},"PeriodicalIF":4.8,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11699492/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142865942","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}