Hospital-at-home (HaH) services (sometimes referred to as virtual wards) are innovative models aimed at delivering hospital-level care within a patient’s home, using a multidisciplinary team approach and remote monitoring technology [1]. Evidence has shown that patients prefer to be cared for at home, and HaH can reduce the need for residential care after patients are discharged from the hospital [2]. There is a significant need to ensure patients are cared for safely and effectively, in the right place, and at the right time. However, caring for patients at home who are acutely unwell comes with its own clinical and operational challenges, particularly for aging populations with increasingly complex health needs [3]. Despite the potential benefits of HaH models, their implementation has been variable, especially the adoption of technology and the consistency in how patients are managed [1,4-9]. Unlike hospital wards, these innovative workflows mean that clinicians are often managing patients remotely with limited direct observation, relying on decisions that are sometimes made on patient-reported intermittent physiological measurements. This can create safety blind spots, making it difficult to identify early deterioration, prioritize caseload, and maintain oversight. Increasing service demands and workforce pressures further challenge clinicians’ abilities to maintain remote situational awareness, requiring approaches that can support and streamline their cognitive workload and case management for delivering timely interventions. Compared with typical clinical settings, there is a greater need for proactive and effective risk assessment, status prediction, and care management rather than reactive care [10].

In the era of advancing digital solutions, there has been a strong emphasis on embracing technology to support streamlined, more efficient health care delivery. In England, the National Health Service (NHS) 10-year health plan focuses on 3 major shifts: hospital to community, sickness to prevention (more proactive care), and analog to digital [11]. Current HaH models use remote patient monitoring (RPM) technologies, and some have virtual dashboards to support decision-making. However, their adoption has been variable. These models primarily present current or historical data, failing to reflect a patient’s live status, which can fluctuate rapidly in acute care. In contrast, a digital twin (DT) is a dynamic virtual representation of a person and/or their environment, and it continuously integrates real-time data to forecast future possibilities and inform patient management decisions [10,12,13]. Rather than just displaying static physiological observations, a DT can simulate how a patient’s condition may evolve over time under different scenarios, enabling the prediction of risk, assessment of intervention options, and prioritization of workflows [10,12,13]. Consequently, DTs have the capability to personalize care for an individual patient and/or setting and offer bidirectional data exchange, predictive simulation, and autonomous decision support functions to support patient management. These capabilities extend above and beyond conventional RPM and dashboards as they are supported by data-driven care planning, leveraging AI and machine learning algorithms. The DT concept involves 5 core architectural components and is presented in Figure 1 [10,13]. The risk prediction capability supports clinicians and service providers in introducing early interventions, leading to improved safety. However, beyond this, DTs can optimize care pathways by integrating patient physiological parameters, home environmental parameters, and service-level operational information. Current evidence points to an expanding body of research [14-16], the development of risk prediction models [17,18], and an increasing application of command centers [10,19] to manage caseloads. Despite this, there is little evidence on the use of DT technology in health care in real-world practice, with many DT applications in health care being in early stages of development and nonexistent in HaH models of care [20]. Building on previous research, we know that the adoption of technology or innovation comes with its own challenges, especially when it comes to human factors and multilevel factors (eg, acceptability, cognitive workload, and human-automation interaction) [21]. As these factors influence sustainability, a fundamental understanding of workforce perspectives is essential, taking into consideration the inherent sociotechnical complexity and the interactions among people, services, and technology. Guided by the Non-adoption, Abandonment, Scale-up, Spread, and Sustainability (NASSS) framework of health and care technologies [21], the aim of this study was to explore the perceptions of health and care professionals and the perceived impact of the concept of a DT-enabled HaH model of care in order to anticipate and plan for the potential adoption or nonadoption of such technology in the future.

This study aimed to explore health and care professionals’ perceptions of a DT-enabled HaH model of care and its perceived impact. Although this technology has not yet been demonstrated in practice, data from participants were used to explore a conceptual model, providing an in-depth understanding of acceptability, benefits, and challenges across multiple levels: individual (patients and staff adopters), meso (organizational), and macro (systemic). The use of the NASSS framework, which is based on a synthesis of multiple theories of technology implementation [21], enabled a rich understanding of multiple sociotechnical factors influencing the potential adoption of DT technology and its integration within existing HaH models of care. The findings showed that there are many uncertainties and multilevel factors that can potentially impact adoption. Some include the current conceptual nature of HaH DTs and the unknowns around interoperability and integration. Furthermore, the potential change to staff routines and the need to learn new skills may involve new structures within organizations to embed DTs, leading to initial apprehension among staff. However, in general, professionals’ views of the concept were positive, and they reported that embedding DTs in practice over time was manageable.

This study highlights key themes to guide recommendations for stakeholder consideration, supporting the implementation of HaH DTs and long-term sustainability. First, the results show that improvements are required from a system-wide perspective to enable successful adoption, scale-up, and spread. These may include digital infrastructure, connectivity, and interoperability with a shared vision centered around the patient. These can enable true person-centered care and holistic insights, and facilitate the predictive capabilities of DTs, if accurate data are inputted. This could facilitate raising awareness across multiple organizations, which can support long-term sustainability. Additionally, supportive infrastructure includes human resources and workforce capacity, which may require remodeling and new ways of working to deal with large amounts of data. There is potential for hybrid roles, working across boundaries, and a level of flexibility. While there is a scarcity of studies around the adoption of DT technology in health care, particularly in a patient’s home, this study advances new knowledge that contributes to and builds upon what is already known about the adoption of existing technologies such as telehealth and remote monitoring. However, this study was conducted in a systematic manner using a robust implementation framework and differs conceptually by considering the potential leverage that DTs offer, such as real-time simulation and predictive analytics. Findings from studies examining telehealth adoption have noted that system-wide improvements and service enhancement or redesign are needed to achieve sustainability [31]. A study by Thomas et al [32], which explored telehealth adoption in Australia using the NASSS framework, similarly found that implementation strategies are needed at multiple levels to support this sociotechnical complexity. The authors noted the importance of raising awareness, as well as clinicians’ desires for flexible hybrid models, dedicated spaces, and structured staff training. The findings broadly correlate with the work of Sugarhood et al [33], who described that innovative technology cannot be seen as “plug and play” but as a complex innovation requiring strengthened linkage between different parts of the system, with input and coordination between people and organizations. This may require extensive changes to systems, structures, and ways of working, focusing on optimizing human-system integration. The authors describe the technology itself as a “hard core” (the unchangeable features of the innovation itself) that requires a “soft periphery” (systems and services essential to full implementation) [33]. Comparably, this “soft periphery” requires “organizational innovativeness” and receptiveness among organizations and leaders, which can vary [33]. Similar to our study, other research has described that technological innovations face challenges such as competing priorities and organizational financial pressures [33].

Second, adapting working models and using advanced technology require robust and ongoing training and support for reassurance, sense-making, and enhancing human-system interaction. Technical support and in-house DT champions were highlighted by participants as potential key facilitators to improve adoptability and provide supportive reassurance for any initial reluctance in the workforce. This training needs to be incorporated into clinicians’ undergraduate and heuristic training to ensure that DTs are used to complement and enhance care and do not replace clinical judgment or lead to overreliance. This aligns with the findings of Sugarhood et al [33], who conducted a key informant interview study exploring telecare technologies as a system innovation. Their study identified that technological innovations require people to think and behave differently, recognizing the importance of staff training and workforce development in adoption. However, while initial successes were noted with designated digital “champions,” these trained staff were found to often move on to new roles, impacting sustainability. Cartledge et al [31] highlighted that continuous support and improvement were key themes impacting adoption, particularly focusing on staff adopters and organizational domains.

Third, this study showed that adoption and implementation require reassuring evidence, suggesting that initial proof-of-concept studies can help identify any initial technical challenges before wider implementation. Participants felt that this would also help clarify which patients would be suitable for DTs, shaping any selection criteria. The recommendations for initial implementation suggest that stakeholder engagement is important throughout, with the involvement of individuals from various levels to cover all domains of the NASSS framework. Some of these may include patients, members of the public, clinicians, allied health practitioners, service leads, commissioners, technology developers and suppliers, digital leads, organization board members, and national policymakers. This is consistent with broader literature on technological adoption, which suggests that trialability (opportunity to experiment before committing to adopt) and observability (visible benefits) are both facilitators for adoption [33]. In the literature, examples for telecare include having a “show flat” or demonstration area attached to a day center for older people [33], and this approach could be considered for DTs in the community [10]. From a different angle, other literature has mentioned that new technology should not just focus on proof-of-concept but rather use co-production from the early stages of design and development to identify how technologies might better fit in with users’ material surroundings, networks, and lifestyles [33]. This is critical as the needs of end users may differ from the needs and capacities of providers [33].

The findings from this study contribute to the growing international literature on DTs, specifically addressing the scarce research on HaH. National drivers advocate for increased use of technology, care closer to home, and strategies for prevention [11]. However, remote monitoring and the use of technology in HaH remain variable and uncertain [7]. Moreover, it has been argued that increased use of technology can support increased use of HaH [7]. Maximizing the potential of technology requires recognizing that technology can automate time-consuming tasks, such as data entry, and safely manage larger patient groups with the use of remote monitoring [7,34]. In their systematic review, Shi et al [35] recognized that current pressures on urgent and emergency services are encouraging innovative care models that use technology, representing a paradigm shift in acute health care delivery [35]. However, available evidence on high-intensity technologies, such as wearables for remote monitoring, remains scarce [8,35].

The findings of this study are consistent with international literature by Serrano et al [36], who similarly reported that most practitioners support the use of remote monitoring technology and perceived that this will become increasingly relevant in the future and advantageous for clinical practice. Serrano et al [36] described common benefits, including continuous monitoring of patients to provide prompt care, efficient communication, and visualization of health trends within a patient’s own environment, allowing personalized care. Furthermore, they reported that the way data are presented via wearables or devices can influence patient understanding and self-care decisions [36]. Some challenges, such as increased workload, data inaccuracy, and higher patient anxiety, were also noted [36]. However, other studies, such as one exploring workforce perceptions on the use of a pregnancy remote monitoring intervention by Lanssens et al [37], found no actual increase in workload when using the technology. Additionally, the systematic review by Serrano et al [36] identified practitioners’ concerns regarding privacy, which was not a key focus found in our data. While these studies are not entirely comparable to studies of HaH DTs, they explored key components of DTs, such as sensing technology (remote monitoring) for data collection, communication in some aspects (telecare), and visualization using dashboards, which are also used in clinical practice. Nonetheless, there remains a lack of literature specifically addressing HaH DTs, particularly regarding predictive analytics and simulation capabilities.

The application of this multilevel integrated framework highlights some of the contextual challenges within complex health care systems, showing that findings do not always fit separately within individual domains. For example, the theme around trust impacts staff and patient adopters as well as the technology domain and value proposition, and it is recognized that trust strengthens with a growing evidence base and time. This also applies to training, which overlaps across multiple domains and has financial impacts that affect the value proposition, organization, wider system, and embedding over time. These domains should not be taken in isolation, but they are useful as a guide to uncover some of the complexities that need to be considered before implementation, ultimately informing action planning and targeted efforts. This builds on previous literature, which highlights that health care systems can be complex, such as the systematic review by Greenhalgh et al [38] that explored the diffusion of innovations in health service organizations. This reinforces that multiple components can impact the diffusion, dissemination, implementation, and sustainability of innovations [38]. This needs to be recognized as an ongoing process [39]. Moreover, different clinical specialities may have varying priorities in a system, and these need to be further understood. This can be embedded into the design and development process, as has been demonstrated by Waddell et al [40], who integrated the NASSS framework into a well-known design process, the double diamond, for their human-centered intervention. This recognizes that stakeholder involvement and an interdisciplinary approach are important throughout the process. In practice, organizations and teams applying DTs in HaH services may wish to use the NASSS-Complexity Assessment Toolkit (NASSS-CAT) [41] to understand their own context-specific complexity across each of the 7 NASSS domains. This also supports stakeholder engagement involving questioning to understand suitability, feasibility, the capacity of organizations, and the readiness of adopters, including patients. Furthermore, the toolkit can be used as an ongoing tool at different points in time.

The use of a diverse sample with a range of participants supported the aims of this study. Participants had varying experiences, both within and outside of HaH and across multiple settings. The included participants had experience as digital leads and information officers, with some having responsibilities for commissioning decisions and others having no prior experience of high-level technology use and having only clinical experience. However, this study may have attracted interest from keen participants who may be viewed as potential early adopters of technology or those who have negative experiences of technology, introducing potential sampling bias. Furthermore, the inclusion of only participants based in the United Kingdom may limit the generalizability of the findings to international contexts. Ultimately, participants’ perceptions were based on theorization of a DT concept rather than first-hand experience of this technology, meaning that further work tailored to specific contexts would be required.

In conclusion, the findings of this qualitative study show that the adoption of innovations, such as HaH DTs, is shaped by multiple factors across several levels. While there is a perception of great value to patient and carer adopters, based on perceived reassurance and reduction in the burden with DTs compared to other technologies, there is a need for patient and public involvement or co-production to strengthen the findings. The use of a robust framework in this study allowed us to uncover some of the challenges with physical, cognitive, and organizational ergonomics that may be present with a HaH-DT concept and assess how stakeholders can build on this moving forward. The findings of this study highlight that system-level improvements and proof-of-concept evaluations support adoption, and digital champions support ongoing sustainability, awareness, and training. This work provides an early foundation for understanding likely adoption, and further research and implementation efforts should prioritize stakeholder involvement to develop clear, actionable plans.