Throughout the course of dementia, emotional health challenges (eg, depression, negative affect, and worry) are common among both persons living with a dementia diagnosis and their informal caregivers (eg, family members and friends) [1,2]. Persons living with dementia and their caregivers present with different emotional needs throughout the course of dementia [3,4]; however, maintaining relationship quality within the context of changing interpersonal dynamics remains an important component of emotional health for both members of the person living with dementia–caregiver dyad [5].

Dyadic research is gaining popularity within the fields of emotional health and dementia care [6,7]. Dyadic research highlights the impact of interpersonal relationships and the transmission of emotional states between members of the person living with dementia–caregiver dyad, which influence the emotional health of each member [8]. For example, perceived relationship strain by the persons living with dementia is associated with increased emotional distress in their caregivers and vice versa [9,10]. Alternatively, positive affective states can be shared among dyad members (eg, laughter), increasing emotional health for both dyad members [11]. Indeed, relationships and the emotional health needs of persons living with dementia and caregivers often change throughout the course of dementia, especially in advanced stages, as cognitive and communication limitations are more common [4,12]. However, despite changes, the need for positive experiences shared between persons living with dementia and caregivers remains common and relates to fewer behavioral and psychiatric symptoms among persons living with dementia and less stress among caregivers [13]. Thus, shared pleasant activities are an important component of emotional health among person living with dementia–caregiver dyads that can be targeted at any stage of illness.

As dementia advances and cognitive impairments become more pronounced, identifying shared activities that can promote emotional health for both persons living with dementia and caregivers may become more challenging [14-16]. Indeed, many dyadic interventions in the context of dementia are concerned with systematic approaches to identifying and engaging in shared activities [17]. For example, the tailored activity program intervention helps dyads identify and engage in personalized shared activities to reduce agitation among persons living with dementia and improve well-being among caregivers [18,19]. However, many of these shared activity interventions rely heavily on caregivers to plan and initiate activities, which may reduce shared positive affective states due to the caregiver-care recipient role, rather than equitable activity engagement [11]. For example, caregivers may find themselves in the role of an “instructor” or “supervisor,” rather than fully engaging in and receiving the emotional benefits of an activity. Therefore, designing interventions that allow both persons living with dementia and caregivers to fully engage in shared activities may create a positive emotional experience for each dyad member, contributing to their emotional health. Technology may aid this endeavor by creating interactive experiences that minimize the need for caregiver initiation (ie, using a technology platform that engages and guides both members of the dyad through an activity, rather than relying on the caregiver to lead the activity). Immersive virtual environment technology (IVET) is one such option for engaging person living with dementia–caregiver dyads in an intervention together.

IVET exists under the umbrella of virtual and augmented reality [20] and is defined as an artificial sensory experience that users actively engage with as if it were a real experience [21]. IVET promotes active engagement between users and computer-simulated environments through multisensory user-driven experiences [13,21,22]. These methods align well with existing approaches to promoting positive affect among persons living with dementia (ie, multisensory stimulation and personalized activities) [23]. IVETs often follow the 3 I’s design framework [24], which suggests that user engagement is achieved through (1) immersion (ie, creating a feeling of presence within the virtual environment), (2) interaction (ie, creating a sense of agency and control over the virtual environment), and (3) imagination (ie, the potential to create new experiences within the virtual environment). In nondementia populations, IVET interventions have demonstrated positive effects on emotional health outcomes, such as positive affect and emotional health (eg, depression and stress) [25]. Importantly, IVET interventions vary in terms of the level of immersion, categorized as low (ie, screen-based intervention), medium (ie, cave automatic virtual environments and surrounding projection screens), or high (ie, head-mounted displays for isolation from physical reality) [25]. In nondementia populations, medium immersion produces the greatest results on emotional health, with little difference observed between low and high immersion [25]. However, given the potential safety risks, such as cybersickness, falls, and agitation associated with higher levels of immersion in the context of dementia [21,26], IVET with low immersion may be suitable for this population. Indeed, most IVET interventions tested within the context of dementia use lower levels of immersion [27].

IVET interventions allow for multiple users to engage jointly (eg, persons living with dementia and their caregivers participate in shared activities) [21]. However, existing interventions have been tested primarily among persons living with dementia and caregivers individually [27-30]. For persons living with dementia, IVET interventions have demonstrated feasibility and usability, primarily targeting cognitive rehabilitation, exercise, or spatial navigation training [27]. For caregivers, IVET interventions have demonstrated feasibility and usability, often focused on caregiver education and skills [31,32]. Fewer IVET interventions have been developed for person living with dementia–caregiver dyads to participate in jointly (ie, dyadically), with a focus on shared positive experiences for emotional health. SENSE-GARDEN is an example of a dyadic IVET intervention, which creates an immersive environment based on preferred autobiographical experiences, such as music, images, videos, and personal media (eg, family photos), for persons living with dementia who have moderate to severe cognitive limitations, allowing them to navigate along with their caregivers. The intervention is designed to support reminiscence between the persons living with dementia and caregivers in the present moment. Notably, caregivers have an active role in personalizing the information in SENSE-GARDEN but receive a lower “active dose” of the intervention (ie, fewer caregiver-focused experiences). Rather, caregiver benefit is conceptualized as arising from positive interactions with persons living with dementia. A clinical trial is currently underway for this intervention [33].

A major challenge to designing dyadic IVET interventions in the context of dementia is addressing cognitively mismatched dyads, which is more common as dementia advances. Promoting shared engagement among person living with dementia–caregiver dyads across the illness spectrum must embrace the 3 I’s of design (ie, immersion, interaction, and imagination) for dyads with potentially different cognitive ability levels. While challenging from a design perspective, if achieved, this type of intervention would be an invaluable opportunity for dyad members to truly engage in shared activities together (ie, not initiated by the caregiver), helping them to briefly abandon the caregiver-care recipient dynamic and, in doing so, derive relational and emotional benefits. These types of more equitable interactions may promote positive affect and interpersonal connection among both persons living with dementia and caregivers, dignity for the persons living with dementia, and positive aspects of caregiving for the caregivers [34]. However, currently, guidelines for designing dyadic IVET interventions to promote shared engagement among person living with dementia–caregiver dyads are lacking, especially for dyads with mismatched cognitive abilities [6,35].

Additional work is needed to guide the development of dyadic IVET interventions throughout the course of dementia, especially in more advanced stages. For this endeavor, user-centered design approaches may be particularly important to ensure that interventions have the potential to be acceptable and feasible for both dyad members at various stages of illness. User-centered design is an iterative approach that centers the user experience early and often throughout the intervention development process [36]. The aim of user-centered design is to maximize the feasibility and acceptability of an intervention [37]. This study followed a multiphasic conceptual model of user-centered design rooted in human factors research, as outlined by Witteman et al [38]; refer to Figure 1. In phase 1, design teams engaged users and other knowledge holders with the goal of understanding user needs, goals, and strengths. In phase 2, design teams iteratively developed and refined an initial intervention prototype with user feedback. In phase 3, design teams observed users’ naturalistic interactions with the initial prototype to inform additional refinements [38]. On the basis of user-centered design principles in the context of dementia [39], we prioritized the following factors across each design phase: understandability (ie, the user understands the intervention functions), learnability (ie, the user is able to easily learn the intervention), usability (ie, the user is able to participate in a task with minimal assistance), and safety (ie, tasks do not inadvertently lead to negative affect or other adverse outcomes).

This study aimed to develop a dyadic IVET intervention to support the emotional health of person living with dementia–caregiver dyads. Our design team represents an academic-industry partnership. In this study, we highlight our multiphasic user-centered design process [38], focused on the early stages of intervention development of a dyadic IVET intervention (phases 1 and 2 of user-centered design). First, we present data from focus groups with clinicians, caregivers, and person living with dementia–caregiver dyads to understand user needs, goals, strengths, and usability and safety concerns (phase 1). Then, we discuss the process of iteratively developing and refining an initial intervention prototype based on user feedback from multiple user workshops with person living with dementia–caregiver dyads (phase 2). The resulting dyadic IVET intervention prototype will be ready for pilot testing in a pilot feasibility trial (phase 3). In this study, by describing phases 1 and 2 of user-centered design, we highlight early design considerations for future teams focused on developing IVET or other novel technologies with person living with dementia–caregiver dyads across stages of dementia.

This study followed a multiphasic user-centered design framework for the early stages of intervention development, as outlined by Witteman et al [38] and with the following dementia-specific considerations incorporated [39]: (1) understanding user needs, goals, strengths, limitations, and safety concerns; (2) iteratively developing and refining an initial prototype with user feedback; and (3) observing users’ naturalistic interactions with the initial prototype. In phase 1, we conducted 5 focus groups with clinicians, caregivers, and person living with dementia–caregiver dyads to identify preferred activity features as well as design considerations for usability (ie, preferred intervention activities and accommodations) and safety in the context of dementia. The study team determined readiness to move to phase 2 after we believed that we understood major safety risks and could generate enough content for mock-ups of an intervention prototype. In phase 2, we conducted 8 beta testing workshops with person living with dementia–caregiver dyads to inform the development and refinement of an initial prototype. We determined readiness to move to phase 3 when usability concerns were sufficiently addressed, as evidenced by minimal usability issues reported by workshop participants. At the conclusion of phase 2, we had a finalized intervention prototype ready for feasibility testing in phase 3, which is not reported in this study. Phases 1 and 2 map onto stage 1A of the National Institutes of Health (NIH) Stage Model of Behavioral Intervention Development (ie, intervention generation, refinement, modification, and adaptation), and phase 3 maps onto stage 1B (ie, pilot testing) [40,41].

Phases 1 and 2 of intervention development were informed by our intervention conceptual model, as suggested by Rochon et al [42]. Specifically, we aimed to develop an IVET intervention that could promote sustained attention, positive emotions, and active engagement (ie, mechanistic targets) among both persons living with dementia and caregivers, leading to increased relationship satisfaction, decreased distress, and reduced agitation among persons living with dementia, as well as decreased burden among caregivers (ie, outcomes).

Procedures were approved by the Massachusetts General Hospital Institutional Review Board (2022P001401). The information reported in this manuscript follows the Guidance for the Reporting of Intervention Development standards [43]. Potential clinician participants provided verbal consent before participating in focus group interviews. Caregivers provided verbal consent or e-consent during screening phone calls. A research assistant administered the University of California, San Diego Brief Assessment of Capacity to Consent Questionnaire (UBACC [44]) to persons living with dementia to determine their ability to provide consent. For those unable to provide consent, the persons living with dementia provided assent, and their caregivers provided surrogate consent to participate. All data were deidentified to protect anonymity. In phase 1, clinicians were compensated US $30, and persons living with dementia and caregivers were compensated US $50 each. In phase 2, persons living with dementia and caregivers were compensated US $50 each for their time.

During phases 1 and 2, this study enrolled three categories of participants: (1) clinicians involved in dementia care, (2) caregivers of persons living with dementia, and (3) person living with dementia–caregiver dyads.

Clinicians were recruited via emails to dementia care clinics within the affiliated health care system. Clinicians were eligible if they (1) self-identified as dementia care providers and (2) were able to participate via Zoom (Zoom Communications). Of the 11 clinicians screened for focus groups, 1 (9%) did not participate due to a scheduling conflict.

Caregivers and dyads were both recruited via research registries (eg, study fliers). Recruitment materials were designed to target caregivers. Interested caregivers responded to the research team to obtain additional information about the study and review eligibility. Caregivers were eligible if they (1) aged >18 years, (2) self-identified as informal caregivers for a person living with dementia (eg, a family member or friend providing support), (3) could participate in English, and (4) had access to a video-enabled device for Zoom participation. Of the 31 caregivers screened for focus groups, 5 (16%) were ineligible due to not identifying as a caregiver of a person living with dementia, 13 (42%) had scheduling conflicts, and 5 (16%) could not be contacted.

For person living with dementia–caregiver dyads, caregivers were targeted using similar recruitment strategies discussed above. Interested caregivers identified and contacted their care recipient with dementia to participate in the study. Then, caregivers and persons living with dementia participated in a screening call together. Using the same eligibility criteria stated in the Participant Recruitment section, caregivers provided verbal consent or e-consent during screening calls. Eligibility criteria for persons living with dementia included (1) having an informal caregiver participating in the study, (2) a self- or caregiver-reported dementia diagnosis, and (3) the ability to participate in an interview conducted in English. The UBACC uses 2 open-ended questions to assess the understanding of informed consent information (eg, “Can you briefly tell me what the purpose of the study is based on what I just read?”); scores range from 0 to 2, with scores <2 indicating lack of capacity to provide consent. The UBACC scores show a strong correlation with global cognition [45], suggesting that lower scores (ie, the inability to provide consent) are indicative of more advanced stages of dementia. Across stages, of the 12 persons living with dementia screened for this study, 8 (67%) scored a 0, a total of 2 (17%) scored 1, and 2 (17%) scored 2 (ie, able to consent independently). Of the 26 caregivers screened for workshops, 1 (4%) was ineligible due to not identifying as a caregiver, 4 (15%) had scheduling conflicts, 3 (12%) declined due to the care partner’s worsening medical status, and 9 (35%) could not be contacted.

Usability was assessed using the System Usability Scale (SUS) [46]. The SUS consists of 10 items evaluating the perceived usability of technological platforms, including websites. The SUS is the most widely used scale to assess usability for technology-based interventions in the context of dementia, but self-report challenges may exist with more advanced cognitive limitations [47]. However, the SUS is primarily applied to persons living with dementia with mild cognitive impairments and lacks psychometric data across the full spectrum of dementia (ie, limited data available in advanced stages of dementia) [47]. Furthermore, the SUS uses 5-point Likert scale for response options [48,49]. Therefore, following best practices to reduce cognitive load and ensure that persons living with dementia could meaningfully participate, we adapted this scale to reflect yes or no response categories [47,50]. This decision increased the likelihood that persons living with dementia in this study could provide feedback independently.

A total of 3 additional study-specific feasibility questions targeted entertainment, safety, and control. These items were adapted from previous technology-based intervention studies with participants with mild cognitive impairment and their caregivers [48] and provided feedback salient to our conceptual model (eg, engagement and positive affect) and design aims (eg, safety). These items were also asked in a yes or no format to increase inclusivity for persons living with dementia.

Activities outlined in Phase 1: Focus Groups and Phase 2: Beta Testing Workshops sections culminated in the development of an initial prototype of a web-based dyadic IVET intervention, referred to as the Toolkit for Experiential Well-Being in Dementia (Isle of TEND). The Isle of TEND website contained a home page (Figure 2) with the ability to navigate to 8 modules, which contained different shared activities; refer to Multimedia Appendix 2 for an overview of modules. Consistent with our conceptual model [42], each module in the Isle of TEND featured a shared activity intended to promote sustained attention, positive emotions, and active engagement with the intervention. The intervention was designed to be used by both persons living with dementia and caregivers together in the home.

This study used RDA as an analytic framework to identify themes from focus groups and workshops. RDA is used to quickly identify themes in qualitative data [51]. Rather than written transcriptions, this technique relies on live coding directly from interviews, such as taking observational notes during interviews or while listening to an audio recording [52-54]. For this study, RDA was chosen to identify specific feedback from participants related to preferences, usability, and safety that could guide prototype development and refinement. By quickly capturing this information, our research team was able to communicate findings back to our design team to iteratively respond to feedback and update the intervention prototype to gather more feedback from users.

Consistent with previous studies using RDA [55], we compiled RDA template notes from 2 to 3 team members (ie, 1 PhD-level clinical psychologist and 2 bachelor-level research assistants) into a single matrix for each focus group. Data from this matrix were used to guide discussions with the design team and informed design decisions toward initial prototype development.

At the conclusion of the study, matrices were combined and used to facilitate thematic analysis of focus groups and workshops separately. Focus group data were analyzed as a single dataset (ie, clinician, caregiver, and dyad focus groups analyzed together), such that we identified themes that emerged across groups. We did not compare themes across populations, as this was not consistent with the aims of this study. Two research assistants (ER and AT) trained in qualitative RDA methods independently reviewed matrices to identify codes. Coders met on multiple occasions during the coding process to resolve discrepancies. Codes were grouped by content to form themes. The thematic structure was iteratively developed and confirmed with a meeting between coders and a clinical psychologist on the study team with experience in dementia care (EP). After a thematic structure was agreed upon, the research assistants (ER and AT) reviewed transcripts to identify salient quotes for each thematic category. Both focus groups and workshops followed the same analytic steps. However, transcripts were only reviewed for focus groups because the data collected during workshops produced fewer exemplar quotes (eg, workshops contained more “thinking aloud” methods or behavioral observation, making data less relevant for exemplar quotations).

Three themes emerged from focus groups with clinicians, caregivers, and dyads: (1) designing flexibly to allow users to tailor the intervention experience to their own environmental context and circumstance, (2) designing with the dyad’s clinical and relational needs in mind, and (3) accounting for illness and aging-related challenges in the design.

In general, users rated the intervention positively (Table 1). On the study-specific feasibility scale, 100% (11/11) of caregivers and 67% (6/9) of persons living with dementia (n=2, 22% were unsure or did not respond) felt that the program was safe to use, 82% (9/11) of caregivers and 100% (9/9) of persons living with dementia felt that the program was entertaining, and 82% (9/11) of caregivers and 89% (8/9) of persons living with dementia felt that the program helped them “feel in charge of their own experience.” On the SUS, 82% (9/11) of caregivers and 89% (8/9) of persons living with dementia thought that the program was “easy to use,” and 73% (8/11) of caregivers and 78% (7/9) of persons living with dementia “felt very confident using the program” (n=1, 11% were unsure or did not respond).

Throughout intervention development, our academic-industry team met weekly to discuss results from focus groups and workshops and translate findings into pragmatic design elements. Often, the research team helped provide the context of findings within the illness context of dementia, as well as guide a systematic approach to intervention design that would prepare the team for a future, scientifically rigorous clinical trial. The industry design team often discussed the possibilities and limitations of an IVET intervention delivered with minimal equipment (ie, via a computer or tablet). These perspectives led to productive discussions about incorporating user feedback into intervention design. Together, this iterative process led to the end product of an IVET intervention ready for initial feasibility testing (NIH stage 1B) that was deemed safe and usable by end users (ie, person living with dementia–caregiver dyads). Feedback and design decisions are provided in Table 2.

Following a multiphasic user-centered design framework with dementia-specific considerations [38], this study used focus groups and user workshops to develop and refine the initial prototype of a dyadic IVET intervention to support the emotional health of person living with dementia–caregiver dyads (ie, the Isle of TEND). The intervention was rooted in our conceptual model of shared activities that can promote sustained attention, positive emotions, and active engagement among both dyad members simultaneously [42]. The Isle of TEND is now ready for feasibility testing, consistent with NIH stage 1B.

This initial intervention development study has several limitations to consider. First, our convenience sample of caregivers, dyads, and clinicians consisted of a majority of White participants. In future pilot testing, we will ensure that participants are more representative of the US population. Second, we did not administer cognitive assessments to describe the level of cognitive impairment of our sample. Therefore, we cannot empirically comment on the cognitive profile of participants with objective measurement, including the severity of cognitive impairment. This study used self-reports from persons living with dementia and caregivers to capture the presence of dementia, and the UBACC provides additional insight into the presence of cognitive limitations inhibiting the ability to provide autonomous consent. Indeed, dementia research must balance participant burden, the reality of difficulties obtaining a dementia diagnosis, and study goals. In line with our broader goal of offering TEND as a free intervention for individuals in community settings, we included participants with varying levels of cognitive ability, as reflected by the fact that only 17% (2/12) of our sample was able to provide independent consent. To circumvent challenges with reporting on quantitative surveys for those with more advanced dementia, our team took multiple steps to support persons living with dementia in reporting their own outcomes, including using binary scales and reading questions aloud. However, these methods may have reduced the validity of our measures, and we still cannot be sure that all items were fully comprehended. Finally, we chose RDA over other forms of qualitative analysis. For this study, prioritizing a quick summary of information and relaying this to our design team consistent with RDA was necessary, but this analytic approach may also be more susceptible to researcher bias.

The result of this multiphasic user-centered design study with an academic-industry design team was the development of an initial dyadic IVET intervention for persons living with dementia and their caregivers to use together at home. On the basis of our conceptual model, we aimed to promote sustained attention, positive emotions, and active engagement in both dyad members to increase relationship satisfaction and reduce psychosocial distress. In this study, we highlight early intervention development with a focus on designing for usability and safety. While some of our design considerations may be helpful for future investigators and design teams focusing on dyads managing dementia, we also recognize that there are gaps that still need to be filled, such as promoting mutual engagement among cognitively mismatched dyads.