Chronic conditions such as cardiovascular disease, diabetes, and cancer require sustained lifestyle changes and self-management, yet traditional care models often provide limited support for long-term behavior change. Digital health technologies, particularly virtual agents, computer-generated characters simulating human-like interactions through verbal and nonverbal cues, offer new ways to provide personalized, scalable, and continuous support. However, the ways in which distinct components within such digital health technologies, including those used in chronic care interventions, are chosen and combined remain underreported. We conducted a systematic scoping review to map how behavior change techniques (BCTs), health data types, and delivery channels are rationalized, combined, and applied in virtual agent-delivered interventions for chronic condition management. The review followed established scoping review frameworks and adhered to PRISMA-ScR reporting guidelines. A search was performed across PubMed, Scopus, PsycInfo, WebofScience, and IEEE Xplore in September 2024. Twenty-one studies met the inclusion criteria. We examined the rationales reported by authors for intervention design, categorized as theory-driven, practice-driven, empirically-driven, mixed, or not explicitly stated. Few studies explained why they selected specific techniques or how health data and delivery channels were intended to interact. Across studies, BCTs were identified but often not explicitly labelled. The most common agent-delivered techniques were self-monitoring, feedback, instruction on how to perform a behavior, and prompts and cues. These techniques were typically supported by subjective self-reports (e.g., symptoms, behaviors), objective data (e.g., step counts, blood pressure), adherence data (e.g., activity completion) and user preference data (e.g., preferred timing of reminders). Delivery channels comprised smartphone or tablet apps. This review provides the first systematic map of how BCT-health data-delivery channel combinations are applied in virtual agent interventions for chronic condition management. It highlights foundational design patterns and reporting gaps, emphasizing the need for transparent, theory-informed reporting to guide future development of adaptive, evidence-based digital health tools.

Shared decision-making (SDM) interventions for patients with chronic obstructive pulmonary disease (COPD) often lack effectiveness because they insufficiently reflect individual preferences. This study examined how patient profiles can inform the interaction design of more tailored SDM support. A sequential mixed-methods design combined focus groups with a digital questionnaire, followed by latent class analysis of responses from a representative COPD sample. Three indicative profiles emerged, characterized by (1) traditional decision-making preferences, (2) pre-informed and active decision-making preferences, and (3) higher use of digital information sources. The profiles differed in preferred delivery modes (static versus digital), information visualization formats (circle diagrams versus stacked bars), and decision-making roles (collaborative versus autonomous). Education, health literacy, general health, and tablet use predicted profile membership. These findings highlight the heterogeneity in user interaction and visualization preferences, demonstrating how profiling approaches can support the design of interactive systems that enable more personalized and digitally empowering decision support.

Background:Knee osteoarthritis is a heterogeneous condition characterized by chronic pain, stiffness, and fatigue that fluctuate rapidly over time. Traditional clinical assessments provide only static diagnoses of disease severity, failing to capture the dynamic, day-to-day symptom variability that impacts patient quality of life. While wearable technologies offer the potential for continuous, high-frequency monitoring, previous reviews have examined general technological interventions for knee osteoarthritis management, yet they lack a specific synthesis of technologies for symptom monitoring. Objective:This study aims to synthesize current research on sensor technologies used for the continuous monitoring of knee osteoarthritis symptoms in free-living or simulated daily environments. Specifically, the review seeks to (1) map sensor modalities to specific symptom domains (biomechanical, physiological, and behavioral); (2) evaluate the alignment between objective sensor metrics and patient-reported outcome measures; and (3) identify gaps in current monitoring paradigms. Methods:A systematic literature search was conducted across PubMed, Embase, Web of Science, and IEEE Xplore. The review followed the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines. Eligibility criteria included studies involving participants with knee osteoarthritis using wearable or portable sensors capable of continuous monitoring (eg, inertial measurement units and electrocardiography) and assessing clinical symptoms (eg, pain, fatigue, and stiffness). Studies relying solely on stationary laboratory equipment (eg, force plates) without a portable component were excluded to ensure relevance to real-world applicability. Data regarding sensor types, sampling frequencies, monitored symptoms, and the statistical association between objective features and subjective symptom severity (key findings) were extracted. Results:A total of 16 studies met the inclusion criteria. The summary constructed from the results revealed a distinct technological saturation: the majority of studies (n=6) used inertial measurement units to quantify biomechanical deficits (eg, gait asymmetry and range of motion), which showed robust correlations with functional limitations. In contrast, there was a notable scarcity of research using physiological sensors (eg, electrocardiography and bioimpedance) to monitor systemic symptoms. Crucially, findings highlighted a significant discrepancy between subjective and objective data, particularly in sleep monitoring, where poor self-reported sleep quality predicted pain exacerbations despite stable objective actigraphy metrics. Furthermore, most systems operated as passive data loggers, with a lack of integration into active feedback loops. Conclusions:Unlike previous reviews focused solely on biomechanics, this study innovatively maps the use of sensors across a multidimensional symptom spectrum, revealing a critical gap in the monitoring of fatigue and physiological stress. The findings suggest that current sensor applications are limited by a lack of integration with subjective patient experiences. Real-world implementation requires a hybrid monitoring paradigm that combines the ecological validity of wearable sensors with the clinical relevance of patient-reported outcomes. This approach paves the way for digital phenotyping and active feedback systems, offering a personalized strategy for managing the complex symptom burden of knee osteoarthritis.

Federated learning is gaining increasing traction, including in healthcare applications. The platform presented in this paper, developed by a multidisciplinary consortium, enables privacy-preserving training of machine learning models to generate predictions for patients with chronic obstructive pulmonary disease and comorbidities. In addition, data synchronization and monitoring are facilitated via the HL7 FHIR standard. The platform includes two front ends: a patient-facing smartphone app and a dashboard designed for healthcare professionals, currently in use at three hospitals in Italy, Estonia, and the Netherlands. Source code, synthetic datasets and fitted ML models will be released and indexed on zenodo.org . Initial ML results obtained from models trained with the platform are discussed. The overall architecture and its implementation in European hospitals is shown in this paper.

Counterfactual explanations (CEs) for multivariate time-series classifiers are often difficult to interpret in domains where experts reason in terms of semantic feature groups rather than individual channels. In rehabilitation movement analysis with multi-sensor inertial measurement units (IMUs), clinicians interpret motion through muscle-group and joint-segment abstractions; yet, most existing counterfactual methods operate at the channel level, producing scattered and biomechanically incoherent explanations. We propose a two-stage framework for group-based counterfactual generation in high-dimensional IMU data. We first show that Shapley-Adaptive (SA) group ranking preserves counterfactual validity but fails to enforce group-level sparsity, motivating the need for explicit group selection. We then introduce Learnable Gate (LG) methods, which incorporate trainable per-group relevance gates jointly optimized with perturbation masks. Experiments on the KneE-PAD rehabilitation dataset demonstrate that LG substantially improves modality-group sparsity compared to the channel-level M-CELS baseline while maintaining or improving validity, temporal smoothness, and generation efficiency. Exercise-specific analyses further show that group-structured counterfactuals yield concise, muscle-level corrective guidance aligned with clinical reasoning. Overall, the proposed framework enhances interpretability without sacrificing counterfactual quality, enabling more actionable explanations for rehabilitation movement analysis.

Chronic low back pain (CLBP) is a leading cause of disability worldwide. Physiotherapy is a common treatment, but its effect on physical functioning is generally modest, particularly for patients with severe complaints (i.e., high levels of disability and pain). Virtual Reality (VR) may complement physiotherapy, yet evidence for its effectiveness remains limited. The aim of this study was to assess the effectiveness and feasibility of a VR intervention integrated within physiotherapy for people with severe CLBP. A cluster-RCT across Dutch physiotherapy practices was conducted. Patients in the control group received 12 weeks of usual care following physiotherapy guidelines. Patients in the intervention group received similar usual care, enhanced with integrated, home-based VR consisting of pain education, exercise therapy, and relaxation modules. The primary outcome was physical functioning at three months. Secondary outcomes included feasibility, pain intensity, and catastrophizing. Analyses were conducted using linear mixed-effect models accounting for clustering by physiotherapy practice. Twenty-five patients participated in the intervention group and seven in the VARIETY control group, instead of the planned sample size of 120 participants. Due to poor recruitment (n = 7), we supplemented the VARIETY control group with 18 historical controls from two comparable trials (total control n = 25), effectiveness analyses are therefore exploratory. Between-group differences were neither statistically significant nor clinically relevant for all outcome measures, compared to the VARIETY control group (e.g., ODI mean difference at three months: -4.80, 95

eHealth interventions can support healthy lifestyle change for preventing and managing cardiometabolic conditions. While most eHealth interventions are designed for the general population, these conditions are more prevalent in people from lower socioeconomic backgrounds. Tailoring interventions to the needs and characteristics of this population can increase adherence and engagement, thereby enhancing the overall intervention effectiveness. However, evidence on how to tailor eHealth interventions to users from low socioeconomic backgrounds remains limited. Therefore, this scoping review examines the tailoring approaches implemented within eHealth lifestyle interventions targeting people with cardiometabolic conditions and low socioeconomic position (SEP). We focus on what is being tailored, the tailoring variables, the algorithms used for tailoring, and how the tailoring approaches are evaluated. Using keywords related to low SEP, cardiometabolic conditions, eHealth interventions, lifestyle, and tailoring, we searched electronic databases including Scopus, Web of Science, PubMed, and PsycINFO. We identified 43 eligible articles, with 27 unique eHealth lifestyle interventions targeting primarily diet and exercise. The literature shows a variety of tailoring approaches, albeit with a trend towards tailoring to socioeconomic factors at the design stage. Most interventions (n = 26/27, 96%) used rule-based algorithms for tailoring, primarily through functions such as feedback selection (n = 17/27, 63%) or variable substitution (n = 16/27, 59%). Although evaluation of tailoring was missing from most studies (n = 15/27, 55%), the importance of sociocultural relevance, appropriate language, and health literacy-sensitive design was highlighted. These findings suggest that while tailoring is present in many interventions, the current approaches remain limited in the facilitating technology and dynamic adaptations to SEP-specific needs. Thus, future research should investigate the application of more advanced, but reproducible tailoring algorithms and rigorously evaluate the impact of different tailoring methods on intervention effectiveness. To synthesize our findings, we assembled a framework that encapsulates the key concepts from our review, in combination with envisioned future work.

Background Remote monitoring technologies hold promise for improving the management of heart failure (HF) by providing continuous data collection and timely interventions. However, scientific data on the quality and clinical relevance of information concerning remote monitoring generated by natural language processing (NLP) models like ChatGPT remains unevaluated. Objective In this exploratory study we investigate whether the NLP model of ChatGPT provides accurate and clinically valuable responses to HF-related queries about remote monitoring. Methods In this study, three experienced HF physicians and two HF nurses participated. The 5 experts were presented with a set of 50 questions regarding remote monitoring and HF management and were asked to score each response generated by ChatGPT 3.5 on a 5-point Likert scale ranging (1 Very poor, 2 Poor, 3 Neutral, 4 Good, 5 Excellent). The scoring criteria included overall quality of the answer, completeness, accuracy, expectations met, and clinical value, See Table. Analysis was performed with IBM SPSS 24. Results Preliminary analysis of the evaluation scores indicated that on average ChatGPT provided neutral to good responses with comparable degrees of quality, completeness, and clinical relevance (see Table (top row mean, bottom row standard deviation). There was agreement between the five experts in all domains aside from clinical value, see Figure. Feedback noted mentioned a lack of depth in the information provided, absence of trial results, repetitive answers and risk of incorrect interpretation by non-experts. Occasionally only information related to the topic was provided without truly answering the question. Discussion Overall performance of ChatGPT as a NLP tool for questions concerning remote monitoring in HF patients was of neutral to good quality, though no erroneous answers were given. Overall impressions were that additional detail in information provided would increase the potential value of NLP tools. A deeper analysis of the 50 questions including answers on best practices, parameters, and ethics, etcetera will be available at the HFSA conference.

BACKGROUND:Virtual reality (VR) has been introduced as a novel intervention in chronic musculoskeletal pain (CMP) rehabilitation. However, much remains unknown about the effectiveness of VR for CMP. The aim of this study was to examine the effectiveness of VR in daily life versus no treatment for people with CMP who were on a waiting list to receive secondary pain treatment. METHODS:This study employed a cluster-randomised, controlled design. The intervention group received a novel VR application that offers pain education and pain management techniques. This home-based, stand-alone, immersive VR intervention was advised to be used daily for four weeks. The control group received no treatment. Primary outcome measure was health-related quality of life at four weeks; secondary outcome measures included various pain-related variables (e.g., pain self-efficacy). Intervention effectiveness was analysed using linear mixed models. RESULTS:Fifty-three participants were included in this trial (mean age: 55.5 (SD: 15)) of which 70% were women. No significant between-group differences were found at four weeks on physical (mean difference (95% CI): 0.039 (-2.810 to 2.889), p = 0.978) or mental (3.590 (-1.640 to 8.819), p = 0.172) health-related quality of life or any of the secondary outcome measures. DISCUSSION:The pain education VR intervention showed no effect in improving outcomes for people with CMP who were on a waiting list, compared to a no-intervention control group. Future research should investigate for which patients, settings, and timing this VR intervention could be beneficial. SIGNIFICANCE STATEMENT:Prior research showed that VR could be beneficial for people with CMP. However, the results of this study showed that VR is ineffective as stand-alone therapy for people with CMP who were on a waiting list to receive secondary pain treatment.

This paper presents a comprehensive reference for an innovative low-contact codesign approach, aimed at mitigating sample bias commonly found in traditional co-design workshops for eHealth technologies. By partnering with a regional newspaper (134 000 readers), we engaged the broader public in the co-design process, to tackle the issue of late-life loneliness. We employed co-design fiction, dilemma-driven, and empathic design methods, integrating these within journalistic content to prompt the reader responses. This initiative attracted 77 responses, including 34 from older adults (65 & thorn; years), 27 of whom shared personal experiences with loneliness. Our findings highlight the potential of lowcontact, co-design via newspapers to foster inclusive participation, overcoming the limitations of conventional workshops, and enabling engagement with a more representative population sample. (c) 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Chronic Obstructive Pulmonary Disease (COPD) is the fourth leading cause of death worldwide and is often accompanied by comorbidities. Patients, due to their health condition, may experience a rapid worsening of symptoms, defined as an exacerbation, which could lead to undesirable hospitalizations or emergency care. To lower this burden, early detection of symptoms as well as health monitoring are crucial. Data from routine clinical visits and follow-ups with the use of questionnaires and self-reported symptoms collected via a dedicated mobile app provide valuable insights that might enable a prompt identification of worsening health conditions. To this end, in this paper we propose a real-time exacerbation detection algorithm based on a paper version of a multimorbid symptom diary (the COPE-III study) developed in the scope of the RE-SAMPLE project. Results on its implementation and use in three hospitals across different countries in Europe are reported along with a discussion on its potential and challenges. Finally, we demonstrate that it detects exacerbation events that are associated with 46% of the emergency accesses and 32% of hospitalizations reported at GEM pilot site.

BackgroundChronic obstructive pulmonary disease (COPD) is a common chronic incurable disease. Treatment of COPD often focuses on symptom management and progression prevention using pharmacological and nonpharmacological therapies (eg, medication, inhaler use, and smoking cessation). Self-management is an important aspect of managing COPD. Self-management interventions are increasingly delivered through eHealth, which may help people with COPD engage in self-management. However, little is known about the actual content of these eHealth interventions. ObjectiveThis literature review aimed to investigate the state-of-the-art eHealth self-management technologies for COPD. More specifically, we aimed to investigate the functionality, modality, technology readiness level, underlying theories of the technology, the positive health dimensions addressed, the target population characteristics (ie, the intended population, the included population, and the actual population), the self-management processes, and behavior change techniques. MethodsA scoping review was performed to answer the proposed research questions. The databases PubMed, Scopus, PsycINFO (via EBSCO), and Wiley were searched for relevant articles. We identified articles published between January 1, 2012, and June 1, 2022, that described eHealth self-management interventions for COPD. Identified articles were screened for eligibility using the web-based software Rayyan.ai. Eligible articles were identified, assessed, and categorized by the reviewers, either directly or through a combination of methods, using Atlas.ti version 9.1.7.0. Thereafter, data were charted accordingly and presented with the purpose of giving an overview of currently available literature while highlighting existing gaps. ResultsA total of 101 eligible articles were included. This review found that most eHealth technologies (91/101, 90.1%) enable patients to self-monitor their symptoms using (smart) measuring devices (39/91, 43%), smartphones (27/91, 30%), or tablets (25/91, 27%). The self-management process of “taking ownership of health needs” (94/101, 93.1%), the behavior change technique of “feedback and monitoring” (88/101, 87%), and the positive health dimension of “bodily functioning” (101/101, 100%) were most often addressed. The inclusion criteria of studies and the actual populations reached show that a subset of people with COPD participate in eHealth studies. ConclusionsThe current body of literature related to eHealth interventions has a strong tendency toward managing the physical aspect of COPD self-management. The necessity to specify inclusion criteria to control variables, combined with the practical challenges of recruiting diverse participants, leads to people with COPD being included in eHealth studies that only represent a subgroup of the whole population. Therefore, future research should be aware of this unintentional blind spot, make efforts to reach the underrepresented population, and address multiple dimensions of the positive health paradigm.

Introduction Chronic Obstructive Pulmonary Disease (COPD) is an incurable chronic disease, and self-management is often used to support patients. Current research often targets clinical aspects, while actual self-management is performed by the patient at home. However, little is known about the patient experience. Objectives This research identifies which self-management strategies people with COPD apply and what the facilitators and barriers are to adopting these. Specific attention is given to the recruitment approach, aiming to increase response rates and the generalizability of the self-management model. Methods A self-management survey developed for people with rheumatic and musculoskeletal diseases (RMDs) was adapted for COPD, pilot-tested, and disseminated via traditional (e.g., via email) and enhanced (e.g., offline support) recruitment routes. Anonymized responses were deductively coded, using the self-management model for RMDs and the model of positive health. Results From 33 respondents, 152 self-management strategies were identified. All strategies could be categorised using the self-management model. ‘Physical activity’ was the most common category. Motivations to start a strategy are mostly derived from the ‘bodily functioning dimension’. Participants reported 122 facilitators and 41 barriers, such as ‘time’ and ‘support’. Passive approaches, in which participants themselves have to sign up, to improve response rates, were not substantial. Conclusion People with COPD perform diverse self-management strategies. These efforts may not always be visible in the clinical setting, as these are often initiated by one’s search journey and thus are additional to standard Healthcare Professionals´ (HCP) advice. Future research should investigate alternative approaches to reach the wider COPD population.

Counterfactual explanations are increasingly proposed as interpretable mechanisms to achieve algorithmic recourse. However, current counterfactual techniques for time series classification are predominantly designed with static data assumptions and focus on generating minimal input perturbations to flip model predictions. This paper argues that such approaches are fundamentally insufficient in clinical recommendation settings, where interventions unfold over time and must be causally plausible and temporally coherent. We advocate for a shift towards counterfactuals that reflect sustained, goal-directed interventions aligned with clinical reasoning and patient-specific dynamics. We identify critical gaps in existing methods that limit their practical applicability, specifically, temporal blind spots and the lack of user-centered considerations in both method design and evaluation metrics. To support our position, we conduct a robustness analysis of several state-of-the-art methods for time series and show that the generated counterfactuals are highly sensitive to stochastic noise. This finding highlights their limited reliability in real-world clinical settings, where minor measurement variations are inevitable. We conclude by calling for methods and evaluation frameworks that go beyond mere prediction changes without considering feasibility or actionability. We emphasize the need for actionable, purpose-driven interventions that are feasible in real-world contexts for the users of such applications.

Accurate physical activity level (PAL) classification could be beneficial for osteoarthritis (OA) management. This study examines the impact of sensor placement and deep learning models on AL classification using Metabolic Equivalent of Task values. The results show that the addition of an ankle sensor (WA) significantly improves the classification of high intensity activities compared to wrist-only configuration (53% to 86.2%). The CNN-LSTM model achieves the highest accuracy (95.09%). Statistical analysis confirms multi-sensor setups outperform single-sensor configurations (p < 0.05). The WA configuration offers a balance between usability and accuracy, making it a cost-effective solution for AL monitoring, particularly in OA management.

Background: Asthma is one of childhood’s most prevalent chronic conditions significantly impacting the quality of life. Current asthma management lacks real-time, objective, and longitudinal monitoring reflected by a high prevalence of uncontrolled asthma. Long-term home monitoring promises to establish new clinical endpoints for timely anticipation. In addition, integrating eHealth interventions holds promise for timely and appropriate medical anticipation for controlling symptoms and preventing asthma exacerbations. Objectives: This study aims to provide a pragmatic study design for gaining insight into longitudinal monitoring, assessing, and comparing eHealth interventions’ short- and long-term effects on improving pediatric asthma care. Design: The CIRCUS study design is a cohort multiple randomized controlled trial (cmRCT) with a dynamic cohort of 300 pediatric asthma patients. Methods: The study gathers observational and patient-reported measurements at set moments including patient characteristics, healthcare utilization, and asthma, clinical, and environmental outcomes. Participants are randomly appointed to the intervention or control group. The effects of the eHealth interventions are assessed and compared to the control group, deploying the CIRCUS outcomes. The participants continue in the CIRCUS cohort after completing the intervention and its follow-up. Results: This study was ethically approved by the Medical Research Ethics Committee (NL85668.100.23) on February 15th, 2024. Discussion: The CIRCUS study can provide a rich and unique dataset that can improve insight into risk factors of asthma exacerbations and yield new clinical endpoints. Furthermore, the effects of eHealth interventions can be assessed and compared with each other both short- and long-term. In addition, patient groups within the patient population can be discerned to tailor eHealth interventions to personalized needs on improving asthma management. Conclusion: In conclusion, CIRCUS can provide valuable clinical data to discern risk factors for asthma exacerbations, identify and compare effective scalable eHealth solutions, and improve pediatric asthma care. Trial registration : The protocol is registered at ClinicalTrials.gov (NCT06278662).

Background:Virtual reality (VR) could possibly alleviate complaints related to chronic musculoskeletal pain (CMP); however, little is known about how it affects pain-related variables on an individual level and how patients experience this intervention. Objective:This study aimed to gain detailed insight into the influence of an at-home VR intervention for pain education and management on pain-related variables, and to explore its feasibility and general experience. Methods:The study applied a single-case experimental design in which an at-home VR intervention was used for 4 weeks by patients with CMP who were on a waiting list for regular pain treatment. Outcome measures included pain-related variables, functioning, and objectively measured outcomes (ie, stress, sleep, and steps). Outcomes were analyzed using data visualization (based on line plots) and statistical methods (ie, Tau-U and reliable change index) on an individual and group level. In addition, a focus group was conducted to assess feasibility and general experience to substantiate findings from the single-case experimental design study. This focus group was analyzed using inductive thematic analysis. Results:A total of 7 participants (female: n=6, 86%) with a median age of 45 (range 31-61) years participated in this study. A dataset with 42 measurement moments was collected with a median of 280 (range 241-315) data points per participant. No statistically significant or clinically relevant differences between the intervention and no-intervention phases were found. Results of the visual analysis of the diary data showed that patients responded differently to the intervention. Results of the focus group with 3 participants showed that the VR intervention was perceived as a feasible and valued additional intervention. Conclusions:Although patients expressed a positive perspective on this VR intervention, it did not seem to influence pain-related outcomes. Individual patients responded differently to the intervention, which implies that this intervention might not be suitable for all patients. Future studies should examine which CMP patients VR is effective for and explore its working mechanisms. In addition, future larger trials should be conducted to complement this study's findings on the effectiveness of this intervention for patients with CMP and whether VR prevents deterioration on the waiting list compared with a control group.