To examine how sleep and movement behaviours, measured on smartphones via ecological momentary assessment (EMA), GPS and accelerometer, impact subsequent daily mood in people with type 2 diabetes (T2D) compared to those without. Sixty-one participants with (n = 32) and without (n = 29) T2D underwent 2 months of smartphone-based data collection through phone sensors (GPS, accelerometer) and EMAs. Daily sleep, movement and mood (happiness, sadness, stress, anger) were assessed. Dynamic structural equation modelling examined the impact of sleep and movement on subsequent mood, adjusted for age, gender and employment status. We found 18 significant within-person effects between smartphone-derived behaviour and subsequent mood, with 17 within-person effects indicating behaviour had a positive effect on mood. For people with and without T2D, higher physical activity, better sleep quality and visiting more locations predicted increased happiness, and higher physical activity predicted lower sadness. However, unique behaviour-mood effects were also found for each group, such as greater actigraphy-derived step count predicting greater anger in people with T2D (0.13 [0.05, 0.2]) but having no effect for those without. Though effects were small, results indicate smartphone-derived behaviour influences daily mood for both people with and without T2D, but that the nuances of these relationships may differ. If daily mood correlates differ between people with and without T2D, digital phenotyping for early detection and intervention may need to be tailored to those with T2D.

Rice stem performs assimilate transport and promises sturdiness due to cell wall structure and composition. However, less is known about the genetic basis of its structural characteristics. In this study, for the first time, the scanning electron microscope (SEM) imaging technique was developed to capture digital phenotypes to assess 18 straw traits collected from the cross-sections of 147 rice accessions. Genome-wide association studies (GWAS) identified 54 significant single-nucleotide polymorphisms (SNPs; integrated into 28 quantitative trait loci) residing in the genic sequences of rice (promoter and coding DNA sequence), and classified into three groups: 1) cell wall-defining genes, 2) cell size-defining genes, and 3) transcription factors. DUF246 and DUF1218, galactose oxidase, mitochondrial Rho GTPase, WUSCHEL-related homeobox 5 and scarecrow-like 9 are the novel genes identified among the 21 candidate genes. These genes may play roles in stem development traits, specifically the distance from the vascular bundle to the end of the parenchymal cells (DVBEPC) and the thickness of the straw cell wall in the protruding part (TSCWP). Post-GWAS analyses showed one significant haplotype on chromosome 4 and 25 significant epistatic interactions. Most notably, nine TF families were repeatedly detected among the significant QTL. Os07g0644300 (XPA-binding protein 2), located in the q7-1 genomic segment and associated with DVBEPC, was found to have a missense mutation. Phenotyping via SEM imaging provides precise genome-phenome association in understanding rice stem cell size and cell wall architecture, which ultimately can define biomass and lodging resistance.

Rapid urbanization and the algorithmically mediated digital environment have been linked to a "digital pandemic" in youth mental health. As Generation Z transitions from play-based to phone-based childhoods, understanding how digital architecture interacts with urban stressors is critical. This review delineates the socio-neurobiological mechanisms underlying this crisis and proposes a comprehensive multi-tiered public health framework for technology-led intervention. Emerging evidence suggests that attention-optimizing algorithms exploit neurodevelopmental vulnerabilities, intensifying negative affect and maladaptive social comparison. Recent studies link digital immersion to circadian disruption, hypothalamic-pituitary-adrenal axis dysregulation, and systemic low-grade inflammation. Urban stressors - including sensory overload and reduced green space - further sensitize the "social brain," creating an evolutionary mismatch that amplifies algorithmic influence and psychological distress. We define the "digital pandemic" as a population-level phenomenon associated with algorithmic pathogenesis, and propose a "Digital Precision Psychiatry" framework that shifts the clinical paradigm from episodic, subjective observation to continuous, objective management. By utilizing digital phenotyping and Just-in-Time Adaptive Interventions (JITAI), this vertically integrated strategy aims to restore bio-psycho-social resilience in youth, turning the digital environment from a source of pathology into a tool for neuroprotection.

Machine learning detection of metabolic syndrome through single-lead electrocardiography: Digital phenotyping of cardiac electrical alterations

Digital phenotyping offers unprecedented opportunities for capturing real-time mental health data through smartphones, yet translating this data into clinically actionable insights remains challenging. While smartphones can generate nearly one million data points per patient per day, health care systems have struggled to incorporate even basic ecological momentary assessment data into routine care workflows. This paper presents a model for clinician-facing data visualizations that can be shared with patients to increase understanding of mental health symptoms and enhance shared decision-making. We describe (1) a participatory design process through which visualizations were cocreated with clinicians; (2) integration of these visualizations into a Digital Navigator-supported (DN) workflow; and (3) a case example illustrating how data visualizations can enhance patient insight and support treatment adjustments. This work was conducted within the Digital Clinic program at Beth Israel Deaconess Medical Center. Fifteen clinicians and 3 clinical supervisors participated in a participatory design process to develop visualizations meeting clinical workflow needs. Data visualizations were integrated into weekly DN sessions following a 3-phase model (guide, refinement, and autonomy) based on self-determination theory. Six visualization types were developed: gauge charts for engagement behaviors, symptom trajectory graphs, correlation matrices linking passive and active data, sleep visualizations, polar/radar charts for multidimensional assessment, and passive-active data integration graphs. A clinical case demonstrates how these visualizations, when delivered through structured DN facilitation, supported patient engagement, behavioral insight, and autonomous self-management across an 8-week treatment program. Thoughtfully designed data visualizations, when developed collaboratively with clinicians and delivered through structured support, can transform digital phenotyping from a technical capability into a practical tool for enhancing engagement, promoting behavioral insight, and supporting self-management in digital mental health care. Future research should examine how this approach affects therapeutic alliance and clinical outcomes across diverse patient populations.

Peripartum depression (PPD) affects ~12-25% of pregnant and postpartum women worldwide, yet routine screening often fails to capture real-time symptom changes. Digital phenotyping (DP), using data from digital devices such as text entries, sleep tracking, physical activity, social media behavior, and ecological momentary assessments, has been proposed as a complementary approach to support the prediction and early identification for PPD. This systematic review (PROSPERO: CRD42023461325) examined 14 studies published between 2014 and March 2025 that explored passive and active DP data across the antenatal and postnatal periods. Most studies employed observational designs and used the Edinburgh Postnatal Depression Scale as the primary outcome. Passive DP data related to sleep and circadian rhythms were frequently associated with depressive symptoms, whereas findings for physical activity were inconsistent. Active DP data, including language features from text entries, mood logs, semi-random ecological momentary assessments, and social media behavior, were often reported as informative, particularly when combined with personal history or self-reported measures. However, considerable variation across study designs, data sources, analytical approaches, and validation strategies limits direct comparison of findings and prevents causal interpretation. Overall, the evidence remains largely exploratory, and findings should be interpreted cautiously pending more rigorous validation.

Machine learning analysis of continuous glucose monitoring identifies a novel dysglycemic phenotype found in most people with cystic fibrosis.

The evaluation of biomechanical parameters in four-beat gaited horses remains limited by the subjectiveness and complexity of current standard methods. Through a deep learning approach, we aimed to infer dissociation % using only acoustic signals. A total of 268 audio samples were extracted from publicly available videos featuring three Brazilian horse breeds (Mangalarga Marchador, Campolina, and Piquira) performing marcha batida and marcha picada. Acoustic features, including root mean square energy (RMS), zero-crossing rate (ZCR), and 13 Mel-frequency cepstral coefficients (MFCCs), were extracted and used to train a long short-term memory (LSTM) neural network. The model accurately predicted the time intervals between successive hoof–ground contacts (R2 = 0.98; MAE = 0.0071), enabling the calculation of the dissociation %. While no significant differences were found between gait types and dissociation %, breed-related differences in both mean hoof–ground contact interval and dissociation were observed, with 8 acoustic features demonstrating discriminative power. Our results suggest that hoof–ground contact patterns can be quantified objectively from audio alone, offering a practical and non-invasive method for gait analysis. The approach holds potential for applications in breed standardization, selection, and digital locomotion phenotyping of horse populations.

Psychosis spectrum disorders exhibit substantial clinical variability, necessitating novel frameworks that transcend traditional symptomatic classification. While polygenic risk accounts for significant heritability, the precise mechanisms translating genetic vulnerability into clinical illness remain elusive. An increasingly compelling hypothesis suggests that immune-related abnormalities link genetic risk with environmental stressors to precipitate neural dysfunction. This review integrates neuroimmunology and computational psychiatry to advance mechanism-driven precision medicine by connecting specific molecular dysfunctions to high-level information processing deficits. We synthesize evidence demonstrating that genetic and environmental risks converge on immune pathways-particularly microglial dysfunction and aberrant synaptic pruning. Functionally, we propose this pathology drives fundamental information processing errors, including maladaptive prior beliefs and reduced sensory precision. Here we highlight the necessity of multi-modal biomarkers and real-time digital phenotyping to stratify patients based on underlying neuro-immune endotypes. Finally, we address the critical challenge of algorithmic bias, emphasizing that proactive, standards-based strategies are required to ensure computational models are equitable and generalizable across diverse global populations. This integrated roadmap offers a path toward a truly personalized, biologically grounded psychiatry.

Monitoring airline pilot mental health: a 3PM framework utilising digital phenotyping and AI

Physical activity and sleep influence fatigue in myasthenia gravis (MG), and digital health technologies (DHT) enable objective monitoring of these behaviors in daily life. Using this approach, we evaluated whether a lifestyle intervention targeting physical activity or sleep hygiene could reduce fatigue in MG. In this three-arm, randomized controlled trial (DIG-MG; NCT05992025), 72 MG patients completed 6 weeks of baseline monitoring with a DHT ring (OURA), followed by 12 weeks of (i) physical activity guidance, (ii) sleep hygiene education, or (iii) observation, and a 6-week follow-up. The primary outcome was the MG Activities of Daily Living (MG-ADL) score 1 week postintervention. Secondary outcomes included Fatigue Severity Scale (FSS) scores; exploratory outcomes were DHT-derived physical activity and sleep parameters. Baseline MG-ADL scores were similar (median: 5.0). Postintervention medians were 4.0 (physical activity), 3.5 (sleep hygiene), and 3.0 (control), with no significant differences (p = 0.073). Clinically meaningful MG-ADL improvement occurred in six, seven, and six participants, respectively. FSS scores showed no group differences (p = 0.992), with clinically relevant improvement in eight participants in each intervention group and five controls. Participants were more physically active than expected: 64.7% exceeded 600 MET-min/week at baseline. DHT adherence was excellent. REM sleep was lower than expected, while deep sleep was preserved. Self-reported data aligned with DHT measurements. Digital lifestyle interventions were feasible and well-accepted but did not improve MG-ADL or FSS in this unusually active population. However, DHT-based monitoring may support individualized follow-up, and reduced REM sleep warrants further investigation as a fatigue-related factor.

Long-term monitoring of behavioral and physiological processes is essential for elucidating complex brain-based phenomena and disorders that develop over extended periods, such as chronic stress, circadian disruption, and metabolic syndromes. Though digital phenotyping is well-established in humans via smart devices, comparable solutions for rodent models remain limited. To address this gap, we present the Digital Homecage (DHC) system, an open-source platform that enables uninterrupted, long-timescale recording of over 20 behavioral metrics in single-housed mice. The DHC integrates video capture, operant task modules, and wheel-running data to achieve sub-second resolution in monitoring behaviors such as actigraphy, sleep, grooming, and food choice over periods spanning weeks. Initial data validate the system's capability to reveal circadian patterns across multiple spontaneous behaviors, consistently reflecting nocturnal activity. Designed for seamless integration with brain recording technologies, the DHC offers unique opportunities for longitudinal analyses of chronic conditions and neuropsychiatric syndromes. Its accessibility and modular design promote community-driven innovation, establishing the DHC as a transformative tool for advancing the study of complex traits in rodent models of brain function and behavior.

Opportunities and Challenges in Using Smartphone Sensor Data to Capture Surgeon Behavior and Well-Being: A Qualitative Analysis of Surgical Residents' Perspectives.

Depression is a serious global mental health illness that causes significant suffering to the individual and social impairment in their lives. Compared to the general population, depression shows a higher prevalence among college students. With recent advancements in digital phenotyping data analysis to infer depressive symptoms, machine learning (ML) techniques have been increasingly employed to indicate behaviors related to potential depressive profiles (PDP). However, despite the growing body of work on ML usage to detect depression, few studies have focused on data preprocessing approaches to handle missing values in datasets that go beyond common data imputation. In this study, we conducted a series of experiments to evaluate the combination of data preprocessing methods and ML algorithms for effectively classifying PDP and non-PDP students using data from the Amive project. The primary challenges were implementing a data processing workflow to address missing values and class imbalance, common issues in digital phenotyping datasets, and selecting algorithms capable of handling such data. The experimental results showed promising outcomes, with individual classification models, including Random Forest, XGBoost, and SVM(rbf), achieving accuracies of 77%, 75%, and 76%, respectively. The best performance was obtained by training on datasets that went through outlier filtering, specifically removing rows with four or more missing values. This combination of data preprocessing approaches and ML algorithms resulted in a Random Forest classification model with the best performance ranging between 77% of accuracy and with mean errors metrics of AUC and MCC above 0.5.

Precision medicine tailors prevention, diagnosis and treatment of cardiometabolic diseases to individual genetic, environmental and lifestyle determinants, with the potential to fundamentally change healthcare. However, low-income and middle-income countries (LMICs) and small island developing states (SIDS) experience severe implementation barriers: inadequate healthcare infrastructure, prohibitive costs, under-representation in genomic datasets and additional SIDS-specific constraints. This Perspective advances three specific contributions beyond generic equity calls. First, it delineates distinct precision medicine pathways for larger LMICs versus SIDS, highlighting SIDS opportunities for regional consortia, shared sequencing and/or biobanking hubs and technological leapfrogging via mobile health platforms and digital phenotyping. Second, it emphasizes practical and high-impact entry points that are financially sustainable. Additionally, it advocates for integrating polygenic risk-based stratification into existing non-communicable disease care pathways rather than establishing separate specialist services. Third, it delineates a staged implementation framework that prioritizes ethical oversight and robust data governance, underscoring the importance of privacy safeguards, data sovereignty, equitable benefit sharing, community consent mechanisms and alignment with the Sustainable Development Goals to minimize associated risks of exploitation. Equitable partnerships between LMICs and high-income countries, expansion of diverse genomic data and community-driven innovation will ensure that precision tools effectively target metabolic phenotypes in LMICs and SIDS while advancing global health equity.

Variability in self-reported depression symptomology and associated behavioral markers in digital phenotyping.

Mental health problems tied to negative affective experiences are common among emerging adults, yet conventional symptom questionnaires provide only distal snapshots of dynamic affective processes that unfold in daily life. Using the novel Meet Pandora smartphone application, we examined how everyday affective experiences and passive markers relate to depressive symptoms (PHQ), anxiety symptoms (GAD), and psychological flourishing in college students ( N = 120; N observations (PHQ/GAD) = 372, N observations (flourishing) = 792). We tested three data channels (self-report, voice-derived features, and behavioral indicators) and disaggregated within-person fluctuations from between-person differences using multilevel models. Across outcomes, the proportion of positive emojis selected to reflect one's affective state emerged as the most consistent signal. On days when participants used a higher-than-usual proportion of positive emojis, they reported lower depression and anxiety, and individuals with higher average positive emoji use reported lower symptoms and higher flourishing. Positive emoji use also moderated change over time in flourishing, such that flourishing increased across the study period only among participants with relatively high positive emoji proportions. In contrast, passive features showed more selective associations. Speech rate (words per minute) was linked to lower symptom burden and higher flourishing in some models, and longer average sleep duration was associated with lower anxiety and higher flourishing. Overall, results highlight the value of separating within- from between-person effects when linking digital markers to mental health and suggest that low-burden indicators of positive affect may be especially informative for scalable temporal monitoring of affective risk and wellbeing in young adults.

Nearly 1 in 4 young adults has a chronic condition, yet many feel well despite their diagnosis. Asymptomatic conditions such as prediabetes and hypertension create a unique vulnerability to digital health misinformation, particularly on platforms where inaccurate content is prevalent. Conventional clinical responses, which often just warn patients about online misinformation, fail to address the underlying drivers of this behavior. This viewpoint proposes a novel disease characteristic–based vulnerability framework to understand this challenge, grounded in established behavioral science theories such as the capability, opportunity, and motivation–behavior model; temporal discounting; and the concept of information voids in infodemiology. We identify a critical “information void” for asymptomatic conditions managed primarily through lifestyle modification. This void, created by the absence of symptomatic feedback combined with delayed clinical biomarker feedback, compels patients to seek information online. Instead of viewing this information seeking as a problematic deviation, we reframe it as a “digital phenotype” indicating a patient’s readiness for behavior change. Through case studies illustrating how this framework applies to specific conditions (prediabetes, nonalcoholic fatty liver disease, and untreated hypertension), we demonstrate its practical utility for clinicians, health systems, and policymakers. Evidence supports a multipronged approach: integrating digital health literacy into clinical encounters, providing curated evidence-based resources, and pursuing strategic institutional engagement in digital spaces. While acknowledging the framework’s deliberate simplification and the need for culturally sensitive adaptation across diverse health care settings, this viewpoint offers a generalizable strategy for engaging with patients’ information needs, helping transform a public health challenge into an opportunity for empowerment.

Background:Relapsed ovarian cancer (ROC) presents significant therapeutic challenges, and complete resection during secondary cytoreductive surgery (SCR) has been associated with improved survival. However, the contribution of the tumor microenvironment (TME), particularly cancer-associated fibroblasts (CAFs), to surgical outcomes remains unclear.Objectives:This study aimed to characterize CAFs heterogeneity in ROC and identify specific CAF subsets associated with immune modulation and surgical prognosis.Design:A multi-platform integrative study combining spatial, single-cell, and transcriptomic analyses to investigate CAF phenotypes and their clinical relevance in ROC.Methods:Multiplex immunohistochemistry and spatial digital phenotyping were performed on 31 ROC samples. Single-cell RNA sequencing (scRNA-seq) was conducted on 11 tumors to define CAF clusters. Transcriptomic meta-analysis across multiple external datasets was used to evaluate prognostic significance. Spatial relationships between CAF subsets and immune cells were analyzed, and antigen-presenting CAF signatures were assessed based on marker co-expression patterns.Results:We identified a predominant S100A4+ CAFs population enriched in the tumor-adjacent stroma, characterized by extracellular matrix remodeling and immune-regulatory gene expression. S100A4+ CAFs displayed closer spatial proximity to T cells and were significantly associated with complete tumor resection (R0) outcomes. Furthermore, a distinct antigen-presenting subset co-expressing CD74 (S100A4+apCAFs) exhibited enhanced interaction with CD4+ T cells and was significantly enriched in R0 patients. Meta-analysis across multiple transcriptomic datasets revealed that high expression of S100A4+apCAFs-related genes correlated with improved overall survival (hazard ratio = 0.81, 95% confidence interval: 0.69–0.95).Conclusion:This study identifies S100A4+ CAFs—particularly the antigen-presenting S100A4+apCAF subset—as key components of the ROC microenvironment linked to favorable immune contexture and surgical outcomes. These findings highlight S100A4+apCAFs as potential prognostic biomarkers and immunomodulatory targets, offering new perspectives for personalized therapeutic strategies in ROC.

Chronic urticaria (CU) is a biologically heterogeneous inflammatory skin disease characterized by recurrent wheals and/or angioedema. Despite its uniform clinical presentation, the disease encompasses diverse clinical phenotypes and molecular endotypes driven by distinct immune mechanisms. Integrating these patterns is essential for transitioning from empirical management toward precision medicine. This review explores the convergence of observable clinical phenotypes and underlying molecular endotypes to refine diagnosis and guide stratified therapeutic strategies. A comprehensive literature review was conducted, focusing on clinical clusters (eg, pediatric, elderly, and ethnic variations), immunological drivers (type I autoallergy and type IIb autoimmunity), and emerging targeted therapies. Linking clinical phenotypes with molecular endotypes enables a more personalized approach to CU management. Future directions include the validation of point-of-care biomarkers and the utilization of digital phenotyping to achieve disease modification and long-term remission.