BackgroundAlthough generative artificial intelligence offers substantial potential benefits in healthcare, negative attitudes and elevated anxiety among nurses may hinder its effective integration into clinical practice. Evidence regarding the psychological impact of generative artificial intelligence on nurses remains limited.ObjectiveThis study examined the relationships among sociodemographic characteristics, anxiety, and attitudes toward generative artificial intelligence among nurses.MethodsA cross-sectional correlational design was employed. Data were collected from 312 hospital nurses using online questionnaires assessing sociodemographic characteristics, attitudes toward artificial intelligence, and artificial intelligence-related anxiety. Data were analyzed using IBM Statistical Package for the Social Sciences (SPSS) Statistics software version 28.ResultsHigher levels of artificial intelligence-related anxiety were associated with less favorable attitudes toward artificial intelligence. Sociodemographic characteristics and anxiety scores collectively explained 49.4% of the total variance in attitudes toward artificial intelligence. Gender, experience with artificial intelligence, use of artificial intelligence in nursing care, awareness of artificial intelligence applications in healthcare, hours spent on the internet, age, and professional experience accounted for 24.7% of the variance in negative attitudes toward generative artificial intelligence.ConclusionAnxiety and experiential factors play a central role in shaping nurses’ attitudes toward generative artificial intelligence. Increasing nurses’ exposure to and awareness of artificial intelligence in nursing practice may reduce anxiety and support its acceptance and appropriate use.
Retron-Powered Approach Enables Genome Editing Across Diverse Bacterial Species
For decades, the ability to precisely rewrite bacterial genomes has been largely confined to a single workhorse organism: Escherichia coli. That limitation has slowed efforts to study pathogens, engineer sustainable biomanufacturing strains, and probe how microbes influence human health. While genome editing tools have transformed eukaryotic biology, most high‑efficiency bacterial editors simply haven’t worked outside E. coli.
A new study from the Gladstone Institutes aims to change that. In a large, nine‑lab collaboration, researchers have translated a retron‑based DNA editing system from E. coli into 14 additional bacterial species spanning three major phyla. The work, published in Nature Biotechnology and titled “Genome editing of phylogenetically distinct bacteria using cross-species retron-mediated recombineering,” demonstrates that retrons, bacterial immune elements that continuously produce short DNA strands, can be engineered into portable genome editing modules the authors call recombitrons. “Recombitrons—a genome editing tool created by pairing modified, donor-producing bacterial retrons with single-stranded binding and annealing proteins—have increased the efficiency of recombineering to install flexible, precise edits in the prokaryotic chromosome,” the authors wrote.
Retrons normally function as part of a viral defense system, generating DNA fragments that help bacteria detect and respond to infection. Seth Shipman, PhD, a Gladstone Investigator and senior author of the study, has spent years repurposing this machinery. “We’ve been easily editing E. coli genomes using retrons for years now, which has substantially increased the pace of our fundamental biology and our molecular technology development,” he said. “But we kept hearing from the broader field, asking when there would be a version of this technology that could be put to work in other bacterial species that matter for the environment, industrial processes, or human health.”
Shipman’s lab previously showed that retrons can act as cellular DNA-making factories, generating the donor strands needed for genome editing. In bacteria, the resulting editing tool built by pairing modified retrons with single‑stranded DNA–binding and annealing proteins is known as a recombitron. Until now, however, functional recombitrons existed only in E. coli.
To test whether the architecture could travel, the team designed a panel of 10 retron-based editing systems and partnered with other labs specializing in diverse bacterial species. “We designed all the molecular parts at Gladstone, then sent them to the collaborators, where they ran the experiment in their labs,” said first author Alejandro González‑Delgado, PhD. Samples were then returned to Gladstone for centralized analysis.
The results show broad functionality. The recombitrons worked in all 15 species tested, including clinically relevant pathogens such as Klebsiella pneumoniae and Pseudomonas aeruginosa, as well as fast‑growing biotechnology strains like Vibrio natriegens and Pseudomonas putida. Editing efficiencies varied widely—from fractions of a percent to more than 90%—but the team demonstrated that modifying retron structure or other system components could boost performance in lower‑efficiency hosts.
“Each retron worked differently in different bacteria,” González‑Delgado noted. “This reinforces why it’s important to have lots of different retrons, so scientists can choose the ones best suited to their favorite bacterial species.”
The study provides a roadmap for expanding genome editing into species that have historically been difficult to engineer. Researchers studying microbial pathogenesis, gut ecology, or industrial bioproduction can now match retron systems to their organism of interest.
“My lab builds molecular technology, and we want these technologies to be used as broadly as possible to uncover new biology and intervene in disease,” Shipman said. “We hope it will continue to spread from here.”
The post Retron-Powered Approach Enables Genome Editing Across Diverse Bacterial Species appeared first on GEN – Genetic Engineering and Biotechnology News.
Prepregnancy Lifestyle Risk Factors in Women Seeking Digital Fertility Services: Cross-Sectional Descriptive Study
<strong>Background:</strong> Approximately 1 in 3 pregnancies in the United States are complicated by one or more adverse pregnancy outcomes. This high prevalence contributes to the elevated rates of maternal and infant mortality in the United States. Modifiable prepregnancy or preconception lifestyle factors have been associated with adverse pregnancy outcomes in observational studies, which underscores the importance of preconception care. <strong>Objective:</strong> This cross-sectional descriptive study aimed to (1) estimate the prevalence of preconception lifestyle risk factors among women seeking services from a digital fertility platform, (2) characterize the study population and present relevant reference data, and (3) examine the distribution of prepregnancy lifestyle scores across demographic and clinical subgroups. <strong>Methods:</strong> The digital health company, Doveras Fertility, has built a prepregnancy digital health platform for individuals and couples seeking to optimize their fertility potential. Targeting users prior to initiation of pregnancy, the platform facilitates the assessment of baseline lifestyle risk factors. This paper reports on 396 adult women who sought the platform’s services in a 1-month period between May and June 2024 by completing a digital fertility questionnaire. Self-reported data were analyzed for 6 healthy prepregnancy lifestyle factors known to be associated with maternal health outcomes in prior observational studies, and each participant was given a composite score between 0 to 6 to represent the number of these healthy behaviors reported. The 6 healthy prepregnancy lifestyle factors include a BMI of 18.5 to 24.9 kg/m<sup>2</sup>, not currently smoking, ≥150 min/week of moderate to vigorous physical activity, healthy eating, no daily alcohol intake, and use of a prenatal multivitamin. <strong>Results:</strong> The study population was racially and ethnically diverse, with a mean age of 32.9 (SD 6.3) years. Most (235/396, 59%) participants received a composite score of 3 factors or fewer, and less than 5% (19/396) scored 6 out of 6. For context, this cohort had higher proportions of participants with unhealthy BMI and dietary patterns than those in the reference data. Regarding fertility, 46% (182/396) met the clinical definition of infertility (≥1 year trying to conceive), with the prevalence of infertility ranging from 16% (3/19) among those with the highest lifestyle scores to 59% (17/29) among those with the lowest. <strong>Conclusions:</strong> Most women seeking services from this digital fertility platform exhibited multiple lifestyle factors that have been previously associated with adverse pregnancy outcomes in observational studies. These results suggest that nearly all survey participants have potential risk factors for adverse maternal outcomes and therefore the potential to adopt at least one improvement in their lifestyle behavior. A digital platform may offer an accessible mechanism for identifying and characterizing preconception risk factors; however, future longitudinal studies are needed to evaluate whether platform-based interventions can effectively support behavior change and improve maternal health outcomes.
Multimodal Depression Detection Through Conversational Interactions with an Emotion-Aware Social Robot: Pilot Study
Background: Depression affects more than 300 million people worldwide and is a leading contributor to the global disease burden. Traditional diagnostic methods, such as structured clinical interviews, are reliable but impractical for frequent or large-scale screening. Self-report tools like the Patient Health Questionnaire-8 (PHQ-8) require disclosure and clinician oversight, limiting accessibility. Recent artificial intelligence–based approaches leverage multimodal behavioral cues (linguistic, acoustic, and visual) for automated depression detection but remain constrained by limited adaptability, scarce annotated data, weak emotional expression in real-world settings, and the high computational cost of deployment of socially assistive robots (SARs). Objective: This study introduces Depression Social Assistant Robot (DEPRESAR)-Fusion, a lightweight multimodal depression detection framework designed for natural interactions with emotion-aware SARs. The objective of this study was to enhance detection accuracy in everyday conversations while addressing the challenges of data scarcity, weak emotional cues, and computational efficiency. Methods: DEPRESAR-Fusion integrates acoustic, linguistic, and visual features with an emotion-aware response module powered by large language models to adapt conversational strategies dynamically. To stimulate richer emotional expression, participants were exposed to emotionally evocative videos before SAR interactions. To overcome data scarcity, we augmented training with (1) public depression-related social media corpora and (2) synthetic samples generated via large language models. The proposed multimodal fusion architecture was evaluated on benchmark clinical datasets for both binary depression classification and PHQ-8 regression tasks. Performance was compared against prior multimodal baselines using root mean square error, mean absolute error, and standard classification metrics. Results: Participants who viewed emotional stimuli before interacting with SARs exhibited significantly higher emotional expressiveness, leading to improved model performance. Regression tasks showed lower root mean square error and mean absolute error, while classification tasks achieved significantly higher accuracy than the nonstimulus condition. DEPRESAR-Fusion outperformed prior multimodal baselines across multiple benchmark datasets, achieving state-of-the-art performance in both binary classification and PHQ-8 regression. The system maintained a lightweight architecture suitable for real-time deployment on SARs. Conclusions: DEPRESAR-Fusion demonstrates that integrating emotion induction, data augmentation, and lightweight multimodal fusion can enable accurate and scalable depression detection in naturalistic SAR interactions. By bridging the gap between structured clinical assessments and everyday conversations, this approach highlights the potential of SAR-based systems as nonintrusive, artificial intelligence–driven tools for proactive mental health support.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/79059e30c7d6ea717d347b5b83aa07e2" />
Obesity Leaves Lasting DNA Methylation Memory in Immune Cells
A new study suggests that obesity leaves a durable molecular imprint on the immune system, one that persists long after weight loss and may continue to influence disease risk. Researchers at the University of Birmingham report that key immune cells retain an “epigenetic memory” of obesity, potentially sustaining inflammation and metabolic dysfunction even after patients return to a healthy weight.
The findings, published in EMBO Reports, provide a mechanistic explanation for a long-standing clinical observation: that individuals who lose weight often remain at elevated risk for conditions such as type 2 diabetes, cardiovascular disease, and certain cancers.
Immune cells retain a “memory” of obesity
The study focuses on CD4+ helper T cells, central regulators of immune coordination. By analyzing patient samples across multiple cohorts, including individuals undergoing pharmacological weight loss, rare genetic obesity syndromes, and lifestyle interventions, the researchers identified persistent epigenetic modifications in these cells.
Specifically, obesity was associated with changes in DNA methylation, a process in which chemical tags are added to DNA and alter gene expression without changing the underlying sequence. These modifications effectively encode a molecular memory of prior metabolic state.
As explained by the authors, these epigenetic marks can persist for years after weight loss. “The findings suggest that short-term weight loss may not immediately reduce the risk of some disease conditions associated with obesity,” said Claudio Mauro, PhD, senior author of the study. Instead, the immune system appears to retain a record of past metabolic stress that continues to influence cellular behavior.
Persistence beyond weight loss
The durability of this imprint is striking. The study estimates that obesity-associated DNA methylation patterns in T cells may persist for five to ten years after successful weight reduction. This suggests that immune remodeling lags far behind metabolic normalization.
Supporting this, the team observed similar patterns across diverse experimental systems, including human clinical samples and mouse models of diet-induced obesity. Together, these data point to a conserved biological mechanism rather than a transient or context-specific effect.
This persistent immune memory may help explain why relapse and long-term complications are common in obesity. As noted by Belinda Nedjai, PhD, of Queen Mary University of London, “the immune system retains a molecular record of past metabolic exposures, which may have implications for long-term disease risk and recovery.”
Disruption of cellular housekeeping and aging
At the functional level, the epigenetic changes identified in T cells appear to disrupt two critical biological processes: autophagy and immune senescence.
Autophagy, the process by which cells degrade and recycle damaged components, is essential for maintaining cellular health. The study suggests that obesity-associated DNA methylation impairs this pathway, reducing the cell’s ability to clear waste and maintain homeostasis.
In parallel, the researchers observed effects on immune aging, or senescence. Dysregulated T cells exhibited features of premature aging, potentially contributing to chronic inflammation and reduced immune resilience.
Together, these alterations could create a persistent pro-disease environment, even after weight loss. This reframes obesity not simply as a reversible metabolic state, but as a condition capable of inducing long-term immune reprogramming.
Implications for treatment strategies
The findings have direct implications for how obesity is managed clinically. If immune dysfunction persists for years after weight loss, then short-term interventions may be insufficient to fully restore health.
Instead, sustained weight maintenance—and potentially additional therapies targeting immune reprogramming—may be required. Mauro noted that “ongoing weight management following loss will see the ‘obesity memory’ slowly fade,” though this process may take years.
The study also points to potential therapeutic strategies. Drugs such as SGLT2 inhibitors, already used in diabetes treatment, may help accelerate the reversal of these epigenetic changes by reducing inflammation and promoting clearance of dysfunctional cells.
Rethinking obesity as a chronic immuno-metabolic disease
Beyond its immediate clinical implications, the study contributes to a broader conceptual shift in how obesity is understood. Rather than being defined solely by excess adiposity, obesity emerges as a condition that induces lasting systemic changes, particularly within the immune system.
As Andy Hogan, PhD, of Maynooth University emphasized, “obesity is a chronic progressive and relapsing disease,” and these findings help explain the biological basis of that persistence.
By identifying an epigenetic “memory” within immune cells, the work highlights a previously underappreciated dimension of metabolic disease: its capacity to reprogram immune function over the long term.
Looking ahead
The discovery of obesity-induced immune memory raises new questions about reversibility and intervention. Can these epigenetic marks be actively erased? And if so, how can therapies be designed to accelerate immune recovery?
Future research will likely focus on targeting these pathways directly, with the aim of restoring normal immune function and reducing long-term disease risk.
For now, the findings underscore a key message: losing weight is only part of the story. Fully reversing the biological impact of obesity may require sustained intervention—not just at the metabolic level, but at the level of the immune system itself.
The post Obesity Leaves Lasting DNA Methylation Memory in Immune Cells appeared first on Inside Precision Medicine.
The Power of Multimodality in Multimodal Large Language Models, Unimodal ChatGPT 5.0, and Human Clinical Experts on a Wound Care Certification Examination: Cross-Sectional Comparative Study
Background: Multimodal large language models (MLLMs) capable of integrating visual and textual information represent a promising advancement for clinical applications requiring image interpretation. Wound care assessment, which demands simultaneous analysis of wound photographs and clinical data, provides an ideal domain to evaluate multimodal vs unimodal artificial intelligence capabilities against human expertise. Objective: This study aims to compare the performance of MLLMs, unimodal ChatGPT 5.0, and human clinical experts on a standardized wound care certification examination. Methods: This cross-sectional comparative study evaluated 3 participant groups on a 25-question wound care certification examination spanning 4 clinical domains (Diagnosis, Treatment, Complication Management, and Wound Subtype Knowledge). Participants included 3 MLLMs (Med-PaLM 2, LLaVA-Med, and BioGPT), 1 unimodal large language model (ChatGPT 5.0), and 4 human clinical experts (general surgeon, wound care nurse, and 2 internal medicine physicians). Statistical analyses included one-way ANOVA with Tukey post hoc tests and domain-specific Kruskal-Wallis comparisons. Results: Human experts achieved the highest accuracy (mean 86%, SD 9.1%), followed by MLLMs (mean 78.7%, SD 12.2%), while ChatGPT 5.0 achieved 64% accuracy, failing the 70% certification threshold. Significant overall group differences were observed (=8.42, =.02, η²=0.74). MLLMs significantly outperformed ChatGPT 5.0 (difference=14.7 percentage points, =.03, Cohen =1.38), with the multimodal advantage most pronounced in visually dependent domains: Diagnosis (81% vs 43%, =.008) and Complication Management (72% vs 50%, =.03). No multimodal advantage was observed for text-based Wound Subtype Knowledge (both 67%). Med-PaLM 2 achieved 92% accuracy, matching that of the wound care nurse, while the general surgeon achieved the highest overall performance (96%). Conclusions: MLLMs demonstrate significant performance advantages over unimodal artificial intelligence in wound care assessment, particularly for visually dependent clinical tasks. While human experts with specialized wound care experience maintain overall superiority, the point estimate of the top-performing MLLM (Med-PaLM 2, 92%) fell within the observed range of human scores; however, the underpowered comparison (power=0.52) and wide CIs preclude definitive conclusions regarding noninferiority or equivalence to human experts. These findings support the potential role of MLLMs as clinical decision-support tools, warranting further adequately powered validation studies.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/96b26bb006d808b607e07c1c5b36f0ca" />
Surgeons’ Perceptions on the Utility of a Conceptual Novel Force Sensor at the Surgeon-Tool Interface: Formative Interview Study
Background: Real-time force feedback is essential in many surgical specialties. While previous research has focused on force measured at the tool-tissue interface, little work has explored the benefits, limitations, or opportunities of measuring force at the surgeon-tool interface. Objective: This study aims to explore scenarios in which surgeons from different medical specialties and experience levels could benefit from receiving feedback on the force exerted at the surgeon-tool (or surgeon-tissue) interface. Methods: Exploratory qualitative research was conducted through interviews with medical practitioners (N=15). This study explored perceptions of a conceptual novel force-sensing surgical glove that could provide real-time feedback in terms of usability, utility, value, and limitations. Opportunities and barriers to implement a sensor of this type in clinical practice were also explored. Participants had experience in anesthetics, dental surgery, plastic and dermatological surgery, general surgery, and obstetrics and gynecology, as these surgical fields all require precise feedback on exerted forces. Results: Participants identified two key areas where a force sensor could yield significant benefits: (1) it could enhance surgical training through objective skill assessment and quantifiable feedback, and (2) it could provide valuable insights into the forces applied during practice, particularly in scenarios where other sensory feedback is masked. Participants appreciated that a sensorized glove that can provide real-time force sensing at the surgeon-tool interface would allow for continued feedback irrespective of the instrument, and integrate seamlessly into their current surgical workflow. Furthermore, as surgeons in some specialisms, for example, dental or obstetrics and gynecology, perform manual tasks, having a sensorized glove would provide feedback in instances where they are physically manipulating tissue. However, participants expressed concerns about accurately defining safe force ranges due to the variability in patients’ anatomical structures and the potential interference with tactile sensation. Conclusions: Surgeons from various clinical practices agreed that force sensing at the surgeon-tool interface could be valuable and provide them with optimal versatility as to when they would adopt force sensing. A sensorized glove could improve decision-making and surgical outcomes when other sources of information guiding force exertion are masked. Conversely, it could be detrimental when the organic information to guide force exertion is distorted when using the sensor. While the choice between interaction modalities is dependent on the accessibility of different senses during surgery, design suggestions as to where sensors are best placed on a sensorized glove are dependent on the instrument used or the type of manual procedure conducted.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/bb3334dc3883c15a56e1f8962075b991" />
RoBuster—Corpus Annotated With Risk of Bias Text Spans in Randomized Controlled Trials in Physiotherapy and Rehabilitation: Corpus Development and Annotation Study
Background: Risk of bias (RoB) assessment of randomized clinical trials (RCTs) is vital to answering systematic review questions accurately. Manual RoB assessment for hundreds of RCTs is a cognitively demanding and lengthy process. Automation has the potential to assist reviewers in rapidly identifying text descriptions in RCTs that indicate potential risks of bias. However, no RoB text span annotated corpus could be used to fine-tune or evaluate large language models (LLMs), and there are no established guidelines for annotating the RoB spans in RCTs. Objective: The revised Cochrane RoB 2 test (RoB 2) tool provides comprehensive guidelines for RoB assessment; however, due to the inherent subjectivity of this tool, it cannot be directly used as RoB annotation guidelines. The study aimed to develop precise RoB text span annotation instructions that could address this subjectivity and thus aid the corpus annotation. Methods: We leveraged RoB 2 guidelines to develop visual instructional placards that serve as annotation guidelines for RoB spans and risk judgments. Expert annotators used these visual placards to annotate a dataset named RoBuster, consisting of 41 full-text RCTs from the domains of physiotherapy and rehabilitation. We report interannotator agreement (IAA) between 2 annotators for text span annotations before and after applying visual instructions on a subset (n=9) of RoBuster. We also provide IAA on bias risk judgments using Cohen κ. Moreover, we used a portion of RoBuster (n=10) to evaluate an LLM using a straightforward evaluation framework. This evaluation aimed to gauge the performance of an LLM (here GPT 3.5) in the challenging task of RoB span extraction and demonstrate the utility of this corpus using a straightforward framework. Results: We present a corpus of 41 RCTs with fine-grained text span annotations comprising more than 28,427 tokens belonging to 22 RoB classes. The IAA at the text span level calculated using the F1 measure varies from 0% to 90%, while Cohen κ for risk judgments ranges between –0.235 and 1.0. Using visual instructions for annotation increases the IAA by more than 17 percentage points. LLM (GPT-3.5) shows promising but varied observed agreements with the expert annotation across the different bias questions. Conclusions: Despite having comprehensive bias assessment guidelines and visual instructional placards, RoB annotation remains a complex task. Using visual placards for bias assessment and annotation enhances IAA compared to cases where visual placards are absent; however, text annotation remains challenging for the subjective questions and the questions for which annotation data are unavailable in RCTs. Similarly, while GPT-3.5 demonstrates effectiveness, its accuracy diminishes with more subjective RoB questions and low information availability.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/5b20788bffffa175b86ced00bb8156f1" />
Frugal-Oriented Information and Communication Technology for Development Framework Toward Low-Cost Digital Maternal Health in Low- and Middle-Income Countries: Quantitative Descriptive Study
Background: The Sustainable Development Goals (SDGs) aim to eradicate poverty and inequality while ensuring that all individuals enjoy good health. Among these, target 3.1 seeks to reduce the global maternal mortality ratio to less than 70 per 100,000 live births. However, progress toward this target has been limited, particularly in low- and middle-income countries (LMICs), where health care delivery remains constrained by limited resources. While digital innovations have increasingly been adopted to improve health care access and service delivery, a significant proportion of populations in LMICs continues to experience inadequate access to essential maternal health services. This gap underscores the need for affordable, sustainable, and contextually appropriate strategies that are cost-effective in improving maternal health outcomes in underserved communities. Objective: This study leverages the principles of frugal innovation and information and communication technologies for development (ICT4D) to propose a frugal-oriented ICT4D framework to deliver low-cost digital maternal health solutions in LMIC settings. The framework seeks to optimize the use of available resources, foster equitable access to maternal health care, and contribute toward achieving SDG 3, particularly target 3.1. Methods: The study was conducted in both rural and urban-poor settings in Kenya using a 2-phased quantitative approach. In phase 1, eight theoretical themes relevant to maternal health uptake were explored. These themes were represented on color-coded sorting cards, which participants ranked according to perceived importance. Phase 2 involved administering structured survey questionnaires to collect empirical data. The study included a total of 32 participants, whose insights provided a foundation for analyzing the significance of contextual factors influencing maternal health service utilization. Results: The weighted scores for 3 of the 8 predetermined theoretical themes—such as resources, information services, and social support programs—emerged as the most influential factors shaping maternal health promotion (N=32). Resources ranked highest (n=6, 18.81%), followed by information services (n=6, 17.99%), while social support programs accounted for 9.64% (n=3) of the overall influence. These findings highlight critical enablers and barriers within the maternal health care landscape and provide a nuanced understanding of contextual dynamics that affect the uptake of maternal health services. The results informed the design of a frugal-oriented ICT4D framework that prioritizes low-cost digital interventions tailored to resource-limited settings. Conclusions: Despite increasing recognition of digital innovations as tools for health care transformation in LMICs, adoption of existing capital-intensive solutions remains low due to financial and infrastructural constraints. This study emphasizes the importance of adopting frugal innovation and ICT4D principles in designing low-cost, scalable digital health interventions to improve access to maternal health care. Implementing such approaches can address resource limitations, enhance maternal health outcomes, and support progress toward SDG 3, particularly target 3.1. The proposed framework provides a foundation for future research and practical interventions aimed at sustainable, equitable maternal health service delivery in LMIC contexts.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/a79f286809b201742b849fb573582897" />
Enhancing Sleep and Mental Health: Longitudinal, Observational, Real-World Study From a Digital Mental Health Platform
Background: Poor sleep is closely linked to mental health challenges and workplace burnout. Mental health and workplace stressors can impair sleep, while good sleep quality supports cognitive and emotional resources to cope with daily challenges. Despite positive outcomes of maintaining good sleep, many people struggle to get enough restorative sleep at night. Given the bidirectional relationship between sleep and mental health, evidence-based digital mental health solutions may offer an accessible and scalable approach to improving sleep quality. Objective: This study examines whether engagement with an employer-sponsored, multimodal digital mental health platform is associated with improvements in sleep quality over time, and whether changes in sleep quality are associated with concurrent changes in mental health and burnout outcomes. Methods: This 12-month prospective, observational study followed working adults who were newly registered to an employer-sponsored digital mental health platform (Modern Health). The platform leveraged technology (mobile and web) to connect employees with comprehensive provider-led and self-guided care through therapy, coaching, on-demand digital resources, and group psychoeducational sessions. Participants [N=578; 61.1% (n=353) women; mean age 33.88, SD 8.73 years; 40.3% (n=233) people of color] completed measures of self-rated sleep quality, depression, anxiety, and burnout (exhaustion, cynicism, and professional efficacy) at baseline and after 3 and 12 months of accessing the platform. Upon registering for the platform, participants were given an initial care recommendation, but could flexibly engage in any combination of services. Participants in this study engaged with at least one care modality, including therapy, coaching, psychoeducation sessions, and self-guided mental health resources. We examined perceived sleep quality and associations with other study variables at baseline, changes in perceived sleep quality over time, and whether changes in sleep quality correlated with concurrent changes in mental health and burnout. Results: At baseline, 42% (243/578) reported poor sleep quality and were more likely to have higher levels of depression, anxiety, and burnout. A generalized linear mixed-effects model showed that each additional month of platform access was related to an increased odds of having good sleep quality by 3.7% (=.02). Linear mixed-effects models found that higher sleep quality over time was associated with lower depression, anxiety, exhaustion, cynicism, and efficacy (all <.001). Among participants reporting poor sleep quality at baseline, 44% (62/141) reported good sleep quality at 12 months. Within this subgroup, paired sample tests showed significant reductions in depression (−48.3%) and anxiety (−38.3%), and increased cynicism, burnout, though cynicism levels remained below the cutoff for high burnout (23.9%; all <.01). Conclusions: Use of an employer-sponsored digital mental health platform was associated with meaningful improvements in self-reported sleep quality over 12 months. These gains were associated with significant reductions in depression, anxiety, and burnout symptoms, highlighting broader well-being benefits of comprehensive mental health care.
