DIGITAL HEALTH最新文献

Objective: We aimed to describe and investigate the factors associated with private internet use for health purposes in the post-pandemic era.

Methods: Data were taken from a quota-based online sample (n = 3270, German adult population aged 18 to 74 years; 47 years on average), with data collection took place at the beginning of 2025. Concerning the private use of the internet for health purposes, three areas were explored (presence and, if applicable, hours per week): researching health issues (e.g. treatments or medications), exchanging views or discussing health issues (e.g. in patient forums), and using telemedicine services (e.g. online consultations).

Results: In total, 60.7% of the participants researched health issues, 20.7% of the participants exchanged views or discussed health issues, and 12.0% of the participants used telemedicine services (e.g. online consultations). Among such individuals privately using the internet for health purposes, the average hours per week for such activities were 1.4 h (SD: 2.0; health issues), 1.9 h (SD: 3.0; exchange views), and 1.8 h (SD: 2.7; telemedicine services). Regressions showed that higher odds of using the internet privately for all three health purposes were significantly associated with younger age, living together: married/partnership, a higher frequency of sports activity, a health-conscious diet, a higher number of chronic conditions, and higher loneliness levels. Some other independent variables such as gender or level of urbanization were partly associated with the outcomes.

Conclusion: Our present study extends our current understanding of using the internet privately for health purposes in Germany. Future longitudinal and cross-country studies are recommended.

Background: Unhealthy Weight Management Behavior (UWMB) refers to harmful weight-control practices that can lead to physical and mental health issues, requiring intervention. This study examines how social media affects UWMB in emerging adult women (18-25 years).

Methods: A systematic review was conducted following PRISMA guidelines. Five databases (PubMed, CINAHL, EMBASE, Web of Science, and Cochrane Library) were searched for studies published between 1 January 2014 and 31 December 2023. Study quality was assessed using JBI critical appraisal tools, and the protocol was registered with PROSPERO (CRD42024542028).

Results: Nine studies were included. Social networking sites (SNS) (n = 6, 66.7%) promoted self-objectification, body comparison, and appearance-focused feedback, contributing to UWMB. Appearance comparisons on SNS triggered body dissatisfaction, exacerbating UWMB. Content communities (n = 4, 44.5%), such as diet/fitness apps, fostered competition and obsession with numbers, further aggravating UWMB. Definitions of UWMB varied across studies, encompassing behaviors like heavy exercise, substance use, surgical methods, calorie-counting obsession, and binge eating, highlighting inconsistencies.

Conclusions: Social media's diverse negative influences on UWMB in emerging adult women are highlighted with the need for clearer definitions and measurements of UWMB. Tailored interventions that address the specific impacts of different social media platforms are essential.

Background: People who have Alzheimer's disease (AD) experience a progressive decline in their neurological function, which leads to mental deterioration and diminished memory abilities, and altered behaviors that affect both patients and their care providers severely. Diagnosis of the disease at an early stage and with precision helps ensure appropriate intervention strategies.

Objectives: Modern artificial intelligence (AI) technology is promising in medical use for imaging and diagnostic work, specifically involving AD detection and classification. This study aims to develop and evaluate an explainable transformer-based framework that leverages Bidirectional-Encoder representations from Image Transformers (BEiT) to automatically classify AD stages from magnetic resonance imaging (MRI) brain scans.

Method: The proposed framework employs BEiT as a feature extractor on a dataset of 8511 MRI brain images categorized into three diagnostic groups (mild, moderate, and no impairment). Class imbalance is addressed through a Wasserstein generative adversarial network with gradient penalty-based oversampling strategy that generates synthetic MRI images for minority classes, and these images are combined with the original scans to form a balanced training set.

Results: The experiments showed outstanding accuracy levels reaching 96%, while the F1-scores indicated 0.94, 1.00, and 0.95 for mild, moderate, and no AD group classifications. Performance evaluation metrics from the study demonstrate strong outcomes with a mean absolute error reaching 0.0727 and Cohen's kappa equaling 0.9451, while Matthews correlation coefficient reached 0.9455 and Hamming loss remained at 0.0365.

Background: mHealth interventions offer significant potential to reduce internalised stigma among men who have sex with men (MSM) living with HIV in low- and middle-income countries (LMICs). These digital tools offer private, accessible and culturally adaptable support to address self-stigmas related to mental illness, HIV and internalised homonegativity.

Objective: This narrative review explores mHealth interventions targeting self-stigma related to mental illness, HIV and internalised homonegativity, using the behavioural intervention technology (BIT) model as a guiding framework.

Design: Narrative review.

Methods: Studies on digital interventions addressing internalised stigma in the context of HIV, mental health and sexual identity were identified and synthesised. The BIT model guided the analysis of intervention content, theoretical underpinnings and technical features.

Results: Most interventions lacked a clear theoretical framework, culturally tailored content, and detailed reporting of behaviour change strategies and technical design-factors limiting scalability and effectiveness.

Conclusion: Future interventions to reduce internalised stigma among MSM living with HIV in LMICs employing mHealth tools should be grounded in theory, culturally relevant messaging, with clearly specified innovative technical features.

Background: Individuals with prehypertension are at risk of developing hypertension, which affects many adults globally. Sustained physical activity (PA) can lower blood pressure, but maintaining long-term behavior change remains difficult. While PA habit formation interventions are promising, they face issues with scalability and accessibility. At the same time, behavior change chatbots have appeared, but their development often lacks systematic methods. Additionally, optimizing large language models (LLMs) to improve chatbot efficiency and reduce costs still needs more research.

Objective: This study introduces HabitBot, an LLM-integrated chatbot designed to foster PA habits in prehypertensive adults. HabitBot leverages LLMs for seamless interactions and integrates multidisciplinary insights, theoretical frameworks, and evidence to enhance the behavior change process.

Methods: HabitBot was developed through a systematic five-phase process: Phase 1, needs assessment via multidisciplinary discussions; Phase 2, literature review to identify relevant behavior change theories; Phase 3, selection of effective behavior change techniques (BCTs); Phase 4, intervention mapping for prototype design; and Phase 5, usability testing and focus group interviews for refinement.

Results: The process led to eight identified user needs and synthesized the Health Action Process Approach with Habit Formation Theory. Twelve effective BCTs were selected. The prototype was developed and refined across six dimensions based on user feedback. Evaluations indicated high usability, with a mean chatbot usability score of 3.84 (SD 0.82).

Conclusion: HabitBot integrates behavior change strategies with advanced LLM technology, representing a novel approach in chronic disease prevention. Future research should assess its long-term impact and generalizability.

Background: Software-based preoperative 3D shoulder arthroplasty implant planning is traditionally done manually and can be time-consuming. Systems like Blueprint^© are now using software features that suggest implant positioning and sizing. This study is a comparative evaluation of preoperative planning software, assessing user interaction efficiency by comparing the time and number of actions required to complete planning tasks with and without automated planning features.

Materials and methods: Real world plannings log files extracted from preoperative software were used to compare the impact of the auto-planning suggestion feature on the time taken and the number of actions to perform shoulder surgical planning. Comparative analyses were performed across the two main groups (with and without auto-planning) and three subgroups divided on surgeon's experience.

Results: A total of 7021 preoperative plannings done by 1018 surgeons were included. The auto-planning group included 65 surgeons with 780 plannings and the Manual group included 953 surgeons with 6241 plannings.The Blueprint® new preoperative auto-planning feature-marketed under the name "BP Assist"-reduced the number of actions during the planning process by 29% compared to manual planning. For the most complex cases, auto-planning reduced the planning time by 25%. The auto-planning feature used by new users resulted in preoperative planning times that were at least as fast-and often faster-than those of high experience users who planned manually, while also requiring fewer actions to complete the planning process.

Conclusions: The integration of an auto-planning feature into preoperative software allowed for a reduction in the number of actions and the time spent during preoperative software-based planning process. Further research is needed to determine whether auto-planning represents a meaningful advancement in shoulder arthroplasty, with potential to support surgical efficiency and clinical decision-making.

Background: With the emergence of artificial intelligence in medical imaging, large language models such as chat generative pre-trained transformer (ChatGPT)-4o have drawn much attention for their potential in diagnostic support. However, their performance in nuclear medicine applications still remains underexplored. In this study, we aimed to evaluate the Taiwan Food and Drug Administration (TFDA)-approved bone scintigraphy (BS platform) and ChatGPT-4o capability to interpret BS images for the detection and localization of bone metastases.

Methods: A total of 52 BS images were analyzed with three interpretation methods: board-certified physicians, ChatGPT-4o multimodal image analysis, and the BS platform. The performance of the interpretations was evaluated with both binary classification and lesion localization of nine predefined anatomical regions. These results were compared to the report of board-certified nuclear medicine physicians, which served as the gold standard in this study.

Results: In binary classification, ChatGPT-4o achieved an accuracy of 84.6%, similar to the performance of the BS platform's accuracy of 82.7%. However, ChatGPT-4o showed lower performance in lesion localization. Its regional precision was 32.5%, and sensitivity was 13.3%, compared to the BS platform's precision of 80.3% and sensitivity of 64.9%.

Conclusion: ChatGPT-4o showed preliminary potential for detecting bone metastases and assisting in structured report drafting, but its limited lesion-localization performance restricts clinical applicability. The BS platform, developed specifically for bone scintigraphy, demonstrated more consistent regional accuracy in this dataset. These results represent an early proof-of-concept comparison, suggesting feasibility for reporting support rather than clinical deployment. Larger, multi-center studies and domain-specific training will be needed to clarify large language models' future role in nuclear medicine.

Objective: To systematically evaluate transfer learning (TL) models for multiclass ocular disease diagnosis and assess their reliability using explainable artificial intelligence (AI).

Methods: Eight pretrained convolutional neural network (CNN) models were evaluated on a public dataset covering cataract, diabetic retinopathy, glaucoma, and normal classes under a unified protocol. Performance was measured using accuracy, precision, recall, and F1-score. Grad-CAM, LIME, and SHAP were used for interpretability, and the Friedman test assessed performance consistency.

Results: Several models achieved near-perfect performance for diabetic retinopathy. DenseNet121 and XceptionNet performed best for cataract detection, while glaucoma showed consistently weaker results, indicating the need for segmentation-based approaches. Despite similar accuracy, explainability revealed substantial differences in model attention. EfficientNetB3 produced the most clinically meaningful visual explanations.

Conclusions: Accuracy alone is insufficient for trustworthy medical AI. Explainable AI is essential for model selection. EfficientNetB3 offers the best balance between performance and interpretability, and glaucoma diagnosis requires more advanced, segmentation-aware pipelines.

Objectives: To develop and validate a smartphone video-based framework using deep learning for quantifying smooth-pursuit abnormalities in Parkinson's disease.

Methods: Smartphone videos (N = 54) from 18 patients with confirmed Parkinson's disease were rigorously annotated to identify 1767 event-level samples (2-second windows), comprising 941 normal and 826 abnormal smooth-pursuit events. Ocular landmarks were extracted using MediaPipe FaceLandmarker. Preprocessing steps included canthus-referenced spatial normalization, Kalman smoothing, and blink filtering. Event samples were encoded as kinematic feature sequences and classified using DP-MDLA Net, a dual-path multi-scale dilated-LSTM attention architecture that fuses convolutional and recurrent representations.

Results: Under a random split regimen for event samples, the framework achieved 96.59% accuracy, 97.50% precision, 95.12% recall, 96.03% F1-score, and an AUC of 0.9939 on the test set (n = 176). Five-fold cross-validation yielded a mean accuracy of 93.04% (SD 1.86%) and a mean AUC of 0.9735 (SD 0.0102). Subject-independent validation (disjoint split by patient) produced an accuracy of 93.57% and an AUC of 0.9693. Ablation without normalization decreased accuracy to 84.09% and AUC to 0.9323, indicating the critical role of landmark-based spatial alignment.

Conclusion: The framework enables robust event-level quantification of smooth-pursuit abnormalities from smartphone video, supporting portable bedside assessment and standardized longitudinal monitoring of Parkinson's disease without specialized equipment.

Background: While Extended Reality (XR) technologies are becoming increasingly prevalent across society, there is a lack of consensus around their utilisation for the management of health and medical procedure anxieties. We undertook an overview of reviews to examine the effectiveness of these technology-based interventions.

Methods: Data were extracted from full-text systematic reviews of patient-directed XR interventions for health and procedural anxiety. Records from the beginning of 2013 until 30 May 2023 were obtained from searches of MEDLINE, Embase, APA PsycINFO and Epistemonikos. Narrative synthesis then examined the consistency, quality and range of eligible research evidence, and reviews were appraised using the AMSTAR-2 tool.

Results: We examined 56 reviews from diverse clinical contexts (35 of which included meta-analysis). Procedural anxieties were most commonly researched, including those relating to needle insertion, acute surgery, dental operations and/or wound care. Other studies focused on more general health anxieties, relating to longer-term treatment and rehabilitation, maternity and chronic conditions. A range of interventions (e.g. distraction- and exposure-based approaches) and technologies (e.g. immersive and non-immersive devices) have been evaluated, although comparisons between different types of interventions are lacking. While XR interventions were generally found to reduce patient anxiety, AMSTAR-2 evaluations highlighted 44/46 of the appraised reviews as low or critically low in quality, and intervention reporting was often lacking in detail.

Conclusions: Evidence in support of XR interventions has not reached maturity and is currently lacking. Therefore, the emerging positive consensus for these techniques should be challenged, and the rationale for adopting such techniques in practice further considered.