Frontiers in digital health最新文献

Introduction: Interstitial/alveolar syndrome (IS) is a condition detectable on lung ultrasound (LUS) that indicates underlying pulmonary or cardiac diseases associated with significant morbidity and increased mortality rates. However, diagnosing IS using LUS can be challenging and time-consuming, and it requires clinical expertise.

Methods: In this study, multiple convolutional neural network (CNN) models were trained as binary classifiers to accurately screen for IS in LUS frames by distinguishing between IS-present and healthy cases. The CNN models were initially pre-trained using a generic image dataset to learn general visual features (ImageNet) and then fine-tuned on our specific dataset of 108 LUS clips from 54 patients (27 healthy and 27 with IS, two clips per patient) to perform a binary classification task. Each clip in the dataset was assessed by a clinical sonographer to determine the presence of IS features or confirm healthy lung status. The dataset was split into training (70%), validation (15%), and testing (15%) sets.

Results: Following the process of fine-tuning, we successfully extracted features from pre-trained DL models. These extracted features were then utilised to train multiple machine learning (ML) classifiers, resulting in significantly improved accuracy in IS classification compared with the individual CNN models. Advanced visual interpretation techniques such as heatmaps based on gradient-weighted class activation mapping (Grad-CAM) and local interpretable model-agnostic explanations (LIME) were implemented to further analyse the outcomes. The best-trained ML model achieved a test accuracy rate of 98.2%, with specificity, recall, precision, and F1 score values above 97.9%.

Conclusion: Our study demonstrates the feasibility of using a pre-trained CNN as a diagnostic tool for IS screening on LUS frames, integrating targeted data filtering, feature extraction, and fusion techniques. The data-filtering technique refines the training dataset by excluding LUS frames that lack IS-related features (e.g., absence of B-lines). Feature fusion combines features learnt from different models or "fused" to enhance overall predictive performance. This study confirms the practicality of using pre-trained CNN models with feature extraction and fusion techniques for screening IS using LUS frames. This represents a noteworthy advancement in improving the efficiency of diagnosis. In the next steps, validation on larger datasets will assess the applicability and robustness of these CNN models in more complex clinical settings.

Background: Prediction models for Type 2 Diabetes Mellitus (T2DM) often rely on biochemical markers such as glycated hemoglobin, fasting glucose, or lipid profiles. While clinically informative, these indicators typically reflect established dysglycemia, limiting their value for early prevention. In contrast, psychosocial stress, sleep disturbance, tobacco use, and dietary quality represent modifiable, non-clinical factors that can be observed long before metabolic abnormalities are clinically detectable. Yet most studies examine these factors in isolation or as additive lifestyle scores, overlooking how their interdependencies reorganize in the preclinical phase. A systems-level approach is therefore needed to capture how disruptions in behavioral coherence signal emerging vulnerability.

Methods: This study develops a dual-analytic framework that integrates Cox proportional hazards models with artificial neural network (ANN) coherence analysis. Using longitudinal data from the UK Biobank (n=15,774; follow-up up to 17 years), we identified non-clinical predictors of incident T2DM and examined how behavioral networks reorganize across health states. Predictors were screened through multivariate survival analysis and mapped into ANN-derived influence matrices to quantify stability, direction, and systemic coherence of relationships among diet, sleep, psychosocial states, and demographics.

Results: Eighteen significant predictors of T2DM onset were identified. Elevated risk was linked to loneliness, psychiatric consultation, emotional distress, insomnia, irregular sleep, tobacco use, and high intake of processed meat, beef, and refined grains. Protective effects were observed for 7-8 h of sleep, oat and muesli consumption, and fermented dairy. ANN analyses revealed a pronounced breakdown of behavioral coherence in T2DM: foods that stabilized mood in healthy individuals became associated with distress, age and BMI lost their anchoring roles, and emotional states emerged as dominant but erratic drivers of diet. These reversals and destabilizations were consistent across model iterations, suggesting robust signatures of preclinical vulnerability.

Conclusion: T2DM risk is better conceptualized as systemic reorganization within behavioral networks rather than the additive effects of isolated factors. By combining survival models with ANN-derived coherence mapping, this study demonstrates that early prediction is possible from modifiable, everyday behaviors without laboratory measures. The framework highlights leverage points for psychologically informed, personalized prevention strategies.

Background: Atrial fibrillation, the world's predominant cardiac arrhythmia, frequently emerges as a complication post-cardiac surgery, leading to serious outcomes like strokes, heart failures, and even death. Due to its often-silent nature, detecting it can be challenging. Smartwatches present a potential solution, offering screening that is more rigorous.

Objective: This prospective observational study sought to assess the Withings Scanwatch's efficacy in identifying postoperative atrial fibrillation.

Methods: After cardiac surgery, patients received a Withings Scanwatch. Over a span of 24 h, both the smartwatch's photoplethysmography sensor and standard telemetry kept track of any atrial fibrillation incidents.

Results: At the end of the study, data from 260 patients was available for assessment. Atrial fibrillation was identified in 32 of these patients, either via telemetry or the smartwatch. Our data revealed a sensitivity of 69.0%, specificity of 98.7%, a positive predictive value of 87.0%, and a negative predictive value of 96.2%.

Conclusions: This clinical study is the first to evaluate the photoplethysmography sensor of the Withings Scanwatch, and it shows that the Scanwatch has high a specificity and moderate sensitivity in detecting postoperative atrial fibrillation. Thus, Scanwatch may support the conventional screening for atrial fibrillation, and potentially reducing complications and costs of atrial fibrillation. Because of lower than expected sensitivity this technology cannot replace conventional monitoring in postoperative patients.

Background: Effective communication is essential for delivering quality healthcare, particularly for individuals with Functional Neurological Disorders (FND), who are often subject to misdiagnosis and stigmatising language that implies symptom fabrication. Variability in communication styles among healthcare professionals may contribute to these challenges, affecting patient understanding and care outcomes.

Methods: This study employed natural language processing (NLP) to analyse clinician-to-clinician and clinician-to-patient communication regarding FND. A total of 869 electronic health records (EHRs) were examined to assess differences in language use and emotional tone across various professionals-specifically, neurologists and psychologists-and different document types, such as discharge summaries and letters to general practitioners (GPs). Latent Dirichlet Allocation (LDA) topic modelling and two complementary sentiment models (VADER and Flair) were applied to the corpus. Sentiment analysis was also applied to evaluate the emotional tone of communications.

Results: Findings revealed distinct communication patterns between neurologists and psychologists. Psychologists frequently used terms related to subjective experiences, such as "trauma" and "awareness," aiming to help patients understand their diagnosis. In contrast, neurologists focused on medicalised narratives, emphasising symptoms like "seizures" and clinical interventions, including assessment ("telemetry") and treatment ("medication"). Sentiment analysis indicated that psychologists tended to use more positive and proactive language, whereas neurologists generally adopted a neutral or cautious tone.

Conclusions: These findings highlight differences in communication styles and emotional tones among professionals involved in FND care. The study underscores the importance of fostering integrated, multidisciplinary care pathways and developing standardised guidelines for clinical terminology in FND to improve communication and patient outcomes. Future research should explore how these communication patterns influence patient experiences and treatment adherence.

Introduction: Cholelithiasis, commonly known as Gallstone disease, occurs when hardened deposits form in the gallbladder or bile ducts. It affects millions of people worldwide and is especially common in women. While many people may not experience any symptoms, symptomatic cases can present with acute cholecystitis and other complications such as pancreatitis and even gallbladder cancer. However, this disease presents a clinical challenge due to its variable symptoms and risk of serious complications. Therefore, early prediction of gallstones is essential for timely intervention.

Method: Thus, our study presents a novel approach for predicting gallstones. In this study, we have presented a Rotational Forest (RoF) classifier optimized using the Bald Eagle Search (BES) algorithm for gallstone prediction based on a tabular dataset. Our research has been conducted across two frameworks: using RoF alone and using RoF with the BES algorithm.

Result: While using RoF alone, an accuracy of 78% and an AUC of 0.867 was obtained using all features. An accuracy of 75.78% and an AUC of 0.860 were obtained for RoF with the BES algorithm using only 17 features. SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) analysis has distinguished CRP, Vitamin D, Obesity, HGB, and BM as the most dominant features.

Discussion: Likewise, we have also compared our work with other novel works and validated the performance of our model for the prediction of gallstones.

Type 2 diabetes mellitus (T2DM) constitutes a rapidly expanding global epidemic whose societal burden is amplified by deep-rooted health inequities. Socio-economic disadvantage, minority ethnicity, low health literacy, and limited access to nutritious food or timely care disproportionately expose under-insured populations to earlier onset, poorer glycaemic control, and higher rates of cardiovascular, renal, and neurocognitive complications. Artificial intelligence (AI) is emerging as a transformative counterforce, capable of mitigating these disparities across the entire care continuum. Early detection and risk prediction have progressed from static clinical scores to dynamic machine-learning (ML) models that integrate multimodal data-electronic health records, genomics, socio-environmental variables, and wearable-derived behavioural signatures-to yield earlier and more accurate identification of high-risk individuals. Complication surveillance is being revolutionised by AI systems that screen for diabetic retinopathy with near-specialist accuracy, forecast renal function decline, and detect pre-ulcerative foot lesions through image-based deep learning, enabling timely, targeted interventions. Convergence with continuous glucose monitoring (CGM) and wearable technologies supports real-time, AI-driven glycaemic forecasting and decision support, while telemedicine platforms extend these benefits to remote or resource-constrained settings. Nevertheless, widespread implementation faces challenges of data heterogeneity, algorithmic bias against minority groups, privacy risks, and the digital divide that could paradoxically widen inequities if left unaddressed. Future directions centre on multimodal large language models, digital-twin simulations for personalised policy testing, and human-in-the-loop governance frameworks that embed ethical oversight, trauma-informed care, and community co-design. Realising AI's societal promise demands coordinated action across patients, clinicians, technologists, and policymakers to ensure solutions are not only clinically effective but also equitable, culturally attuned, and economically sustainable.

Background: The digital divide in health has rapidly expanded during and after the COVID-19 pandemic, with fragmented understanding and an unclear implementation process, for the formal integration of digital health into the healthcare system, which challenges actionable policy development.

Methods: This critical interpretive synthesis (CIS) of the literature aimed to capture the complexity of the digital divide in health. This began with a scoping review of literature published between 2013 and 2023 describing the digital divide in health within the WHO's European Region, in Web of Science, Medline (via Ovid), PsycInfo (via Ovid), and Sociological Abstract (via ProQuest). Three sets of two reviewers independently conducted the selection, and all contributed to the synthesis process.

Results: Of 4,967 original articles identified, 49 articles were included for review. Results revealed a synthesizing argument that the digital divide should be considered as more of a dynamic, entangled, and reciprocal collection of "areas" of phenomenon affecting service users, rather than "levels". Results describe the three synthetic constructs that describe this synthesizing argument.

Conclusion: Findings suggest that digital health solutions should respectfully consider the pace of human healing, long-term user engagement and adaptability. We call for the importance of inter- and multidisciplinary collaboration to ensure effective and context-sensitive implementation in future studies.

Purpose: The social acceptance of health technologies is crucial for the effectiveness and sustainability of healthcare systems amid the demographic change. However, patients' acceptance, which shapes technology use and compliance, is still insufficiently understood.

Methods: In this study, we explore how perceived risks and perceived benefits relate to attributed value as a proxy for social acceptance. Unlike most studies that focus on individual technologies, we measure public perception of 20 very different types of health technologies-ranging from plaster cast and x-Ray to insulin pumps, bionic limbs, and mRNA vaccines. Through an online survey utilizing a convenience sample of 193 participants from Germany and Bulgaria, we assessed perceived risks, benefits, and overall value attributed to these technologies. The study presents a visual mapping of the technologies and investigates the individual and technology-related factors shaping these perceptions.

Results: The findings suggest that perceived benefit is the strongest predictor for overall value (β = +0.886), while perceived risk plays a significant, but much smaller role (β = -0.133). Together, both factors explain 95% of the variance in overall attributed value (95%, R ²  = .959). Further, individual differences, such as prior care experience and trust in physicians, significantly influences the perceptions of health technologies.

Conclusion: We conclude with recommendations for effectively communicating the benefits and risks of health technologies to the public, mitigating biases, and enhancing social acceptance and integration into healthcare systems.

Objectives: Artificial intelligence (AI) has shown increasing promise is orthopedic medicine. However, its role in postoperative rehabilitation remains insufficiently synthesized, particularly when rehabilitation is viewed as a continuous and dynamic care process. This scoping review aims to systematically map current AI applications in postoperative orthopedic rehabilitation, indentify prevailing application patterns and evidence gaps, and clarify their clinical and nursing implications.

Methods: This scoping review was conducted following the methodological framework by Arksey and O'Malley. A comprehensive literature search was conducted in PubMed, CINAHL Complete, The Cochrane Library, Web of Science, Embase, Scopus, IEEE Xplore, SinoMed, China National Knowledge Infrastructure (CNKI), and the WanFang Database for studies published between March 2020 and March 2025. Data extraction and descriptive synthesis were performed on all included studies.

Results: A total of 38 articles were included in this review, encompassing 3 core AI technologies, namely machine learning (ML), natural language processing (NLP), and expert systems (ES). These technologies were mainly applied in patients undergoing joint replacement, fracture repair, and spinal surgery, with the main application scenarios focusing on risk prediction, dynamic feedback, and rehabilitation monitoring. Notably, most studies focused on short-term predictive outcomes, while limited evidence addressed AI-supported intervention adjustment, nursing decision support, or long-term functional recovery.

Conclusion: This review highlights that, despite rapid technological progress, AI in postoperative orthopedic rehabilitation remains largely predictive rather than interventional. The novelty of this review lies in its stage-oriented synthesis of AI applications across the rehabilitation continuum, revealing a critical gap between data-driven prediction and clinically actionable rehabilitation support. Future research should prioritize high-quality, longitudinal studies and shift toward AI-enabled preventive and adaptive rehabilitation strategies to facilitate meaningful clinical translation.