Family Caregiver Experiences of Using a Mobile App for Music-based Training to Support Dementia Care

2.1 Technology Design for Dementia Care

HCI research in dementia promotes a critical view to designing technologies for the variety of experiences associated with dementia [41]. Research has examined specific challenges associated with dementia, such as supporting memory [59], or addressing wandering [80], alongside more psychosocial aspects. For example, research has investigated technologies supporting communication [73], and how conversational agents can be used to empower caregivers and people living with dementia [88]. Other research has examined using virtual reality to provide tailored experiences for people living with dementia [31], as well as using technology to explore therapeutic ideas like reminiscence therapy [10, 77]. While technology can evidently address a variety of challenges and support the psychosocial needs of people living with dementia, a focus on the potential of arts-informed technologies, as well as the need for digital tools to address complex and changing needs, has been brought to the forefront [22, 41].

Several studies have investigated technologies in the residential aged care setting, including supporting engagement and stimulation via large, wall-sized displays [62], interactive media systems to trigger reminiscence [72], and mobile applications to promote shared experiences between visitors and residents [49]. Such technologies show promise, yet there is significantly less research trialling technologies in the community care context, where people with dementia are cared for by family members at home. HCI research has flagged opportunities for designing for the home context of people with dementia, addressing aspects of wellbeing and quality of life at home [11, 33], as well as the importance of designing in a way that is adaptable to changing needs [73]. The work highlights the complexity of needs of people living with dementia, and considers the many individual and social factors that play a role in adoption of a new technology in this context [36, 77]. As such, there remains a need to consider the needs of people with dementia and their caregivers residing in the community, and the complex factors that play a role in the adoption of new technology-based tools.

2.2 Technology for Interpersonal Dementia Care

It is well established that dementia has a significant impact on both the people living with dementia and their family caregivers [6, 9, 17]. Chen et al [12] discussed the importance of focusing not just on care recipients, but also on caregivers, addressing the burdens that often impair caregiver health and wellness. Caregivers often draw on a range of support and information via digital technologies [27, 63, 74], and using these platforms can be helpful for this stressed group [42]. Dementia care technologies can be classified as used either ‘by’, ‘with’, or ‘on’ a person living with dementia [29]. Technology solutions used ‘with’ a person living with dementia may have a greater impact on quality of life, with potential to increase meaningful conversations and sense of belonging [61]. Technologies can be used to support social interaction between people living with dementia and their caregivers, serving as an ice-breaker if initiating and maintaining conversation becomes challenging over time [32]. Combining a focus on both caregivers and the people they care for, technologies that foster agency and provide meaningful interactions between family caregivers and people living with dementia at home offer unique opportunities for both groups to receive benefits [33]. Considering this dyadic relationship for family caregivers is a key avenue for developing caregiver support services [86]. Offering opportunities for connection and training may greatly add to the lives of both caregivers and the people they care for. While prior research in HCI has focused on the needs of people living with dementia, in this study we aimed to support the use of music in care practices, and therefore primarily investigated caregivers’ experiences in the context of the care dyad. We aimed to extend existing literature by examining how a music-based technology can be designed to support the complex needs of both caregivers and people living with dementia.

2.3 Designing for Music-Supported Dementia Care

Music is a powerful non-pharmacological tool that can provide several benefits for people living with dementia. These benefits include supporting the management of symptoms, such as agitation and depression, improving quality of life, and supporting interpersonal connection between people with dementia and their caregivers [1, 13, 20, 45, 67, 69, 75]. Musical experiences exist on a continuum from recreational to therapeutic, with music therapy being the professional use of music as an intervention by a trained therapist [16]. While musical experiences across this continuum can be valuable, practices informed by music therapy can be more strategically targeted at specific care needs for people living with dementia, such as reducing agitation, promoting relaxation, and stimulating reminiscence [16, 75]. As trained therapists are not always available, caregiver training programs informed by music therapy have received attention as a way to assist with providing quality informal care in the home setting [1]. Such caregiver training programs can be described as indirect music therapy, where a therapist may train caregivers to deliver music interventions [44]. There is also potential for technological tools to support the independent use of music therapeutically at home [11].

In HCI, researchers have trialled a range of technologies to support music-based and sound-based experiences, both in the home context and in formal care contexts. In formal care, mobile applications have been designed to facilitate collaborative music performance [24, 37], and experience-centred, interactive music technologies have been used to promote an enjoyable, active form of engagement and shared movement [46, 47, 48]. In another example, Hodge and colleagues [31] designed a tailored virtual reality experience to enable a person living with dementia to re-engage with the experience of attending a concert. In the home setting, one study evaluated an interactive music exergame with people living with dementia and their caregivers (both professional and informal). Participants found the game valuable, in terms of both social and physical effects, especially when integrated into participants’ daily routines [77]. Another study implemented a TV-based system with a tangible interface for music-supported exercises in the home environment, supporting engagement for people living with dementia and providing enjoyment [35]. Other research has focused on designing interactive music systems with and for people living with dementia [51]. In the home context, Houben et al. [33] co-designed interactive sound players with people living with dementia and their partners, which enabled listening to gentle yet meaningful everyday sounds, such birdsong or sounds of the seaside. These prototypes fostered agency in exploring familiar sounds for people living with dementia and facilitated connection and pleasurable moments at home. Outside HCI, research has considered how listening to music playlists can improve quality of life, further highlighting the important role that technology can play in supporting music-based interventions [28].

These findings suggest that designing music-based technologies for people living with dementia and their caregivers may provide much needed support. Much of this work emphasises the importance of design with and for caregivers and people living with dementia, and demonstrates that music interventions need to be individualised; that is, designed to connect with the user’s musical preferences and memories [66]. As such, new music technologies that are easy to integrate into routines, and add to caregivers’ existing strategies, may be well-placed to provide enrichment and address specific challenges faced by people living with dementia and their caregivers. We aim to address this by examining participants’ experiences with a prototype of a mobile application designed to provide music therapy training and example playlists, the MATCH prototype.

Figure 1:

View Figure

3.1 Design

MATCH was designed over four phases. Phase 1 involved scoping the problem. Phase 2 involved developing key design requirements based on prior work investigating training to support caregivers in using music therapeutically at home [11]. In Phase 3 researchers developed a conceptual design of the first version of the app. Phase 4 involved three cycles of iterative design and evaluation, to understand the needs of people living with dementia and their caregivers at home. This was guided by a survey, 5 focus groups, and usability testing [78]. The survey and focus groups led to an understanding of how caregivers and people living with dementia already use music technologies in their everyday care practices, such as regulating mood, providing joy, facilitating social interaction and connection, encouraging reminiscence, providing continuity before and after diagnosis, and making caregiving easier. Finally, usability testing with 3 dyads and 4 caregivers, led to further refinement of the app. Additionally, Thompson et al. [71] validated the training content with music therapy experts and caregivers who previously experienced music therapy interventions, showing the content was relevant, and had high content and face validity.

3.2 Functionality

The current prototype of the MATCH app was available for participants using either iOS or Android, and was compatible with both mobile phones and tablets. Aimed at family caregivers, the app prototype contains modules to train caregivers how to use music for relaxation, attunement (sensitively and musically responding to a person’s expression to tune in empathically), reminiscence, supporting personal care, and movement. These modules are designed to illustrate how music can be used to foster meaningful connections within families affected by dementia, to promote safe and engaging physical activity, decrease agitation and regulate mood during periods of stress, guide families to use music to stimulate reminiscence, while also supporting family caregivers to complete important care activities such as showering and dressing. The modules highlight the types of music and music-based activities that are best used to meet different care needs.

Music therapists designed the training modules informed by music therapy principles, aiming to support caregivers and people living with dementia to use these music strategies independently (that is, without a therapist present). Figure 1 shows an example module page and a playlist creation page. Content is delivered via the app through a series of scripted videos and case studies modelling use of music-based strategies, safety checklists, and a section for problem-solving unanticipated reactions and suggested responses. For example, within the module on Music Attunement, videos guide the caregiver through how to identify energy levels, select appropriate music, adapt music, and apply attunement to caregiving. Supplementary table 1 lists items contained in each module. The ‘General Tips’ and ‘When things go wrong’ sections provide additional information. For example, they provide further advice on singing, and what to do when the person living with dementia is disengaged. Upon first using the app, caregivers are guided through an introduction and information on the science behind music, and then can proceed through the individual modules.

Alongside the training content, the current MATCH prototype interfaces with Spotify for the purpose of playing music. This allowed access to example playlists suggested by music therapists, as well as functionality for caregivers to build their own personalised playlists aligning with the content of each module within the app. Suggested playlists were intended to give caregivers ideas about how to structure a playlist, what types of songs might be helpful for specific therapeutic goals, and ideas for songs to try if they were not sure where to start. These were based on music therapy principles of music known to be effective for specific goals, and tailored to an assumed age-range of participants. For example, the playlist for relaxation included music that was slow, with minimal variation and no lyrics. For other categories, playlists included music with appropriate tempo, lyrical content, and/or from decades likely to be during the ‘reminiscence bump’ [38, 58, 60] of anticipated participants. The purpose of these playlists was to demonstrate how to select and order music for the playlist, and to provide ideas of tracks that may be helpful for situations where caregivers were unsure of their loved one’s musical preferences. Training modules emphasised that these were samples, and that participants should create similar playlists based on preferred music. Within the sample playlists, there was a function for users to add a song to their own personalised playlists.

4.1 Participants and Recruitment

We recruited care dyads, comprising a family caregiver and a person living with dementia or mild cognitive impairment, who were interested in trialling MATCH. Caregiver recruitment was primarily through StepUp (an online platform that connects volunteer research participants with dementia researchers in Australia), and from a database of people who registered interest for related research opportunities at our university. Additional recruitment occurred via a Facebook advertisement, our research project website, our University’s internal staff newsletter, and emails distributed to community groups known to the researchers. To comply with inclusion criteria, the person with dementia needed to be cohabiting or living in close proximity with the primary caregiver and required a formal diagnosis of mild cognitive impairment or dementia. Criteria for exclusion were limited access to the internet or a mobile device for the app and hearing impairment that is not overcome by hearing support. Caregivers also needed sufficient English to use the app, which is presently only available in English, but will be translated into other languages in future iterations. Potential participants were contacted over the phone for initial screening against inclusion and exclusion criteria, as well as availability to participate in the study.

Recruited participants consisted of 17 dyads, with 13 dyads who completed the full 8-week trial. Table 1 shows participants’ demographic details. One dyad withdrew before mid-trial (week 4) due to hospitalisation of the care recipient, and three others withdrew between the mid-point and end of the trial, due to caregiver illness, change in caregiver circumstances, with the fourth dyad lost to follow up, unable to be contacted. Three of the four participants who withdrew consented to us retaining their existing data. For the 17 dyads, participants living with dementia ranged in age from 62 to 93 years (M = 79, SD = 8.4), with 11 women and 6 men. Caregivers ranged in age from 31 to 77 years (M = 59.5, SD = 13.2), with 14 women and 3 men. Among caregivers, 9 were spouses or partners, and 8 were adult children or grandchildren. The majority of participants living with dementia were diagnosed with Alzheimer’s disease (8 participants), while 2 had vascular dementia, 5 had mixed dementia or other diagnoses, and 2 experienced mild cognitive impairment. Three participants had younger onset dementia. Our target population was people residing at home, therefore likely to be in early- to mid-stages of dementia, as those with advanced dementia are often cared for in long-term care homes. Nevertheless, we did not set strict criteria or assess ’stages’ of dementia, as music therapy can have benefits across the trajectory of dementia [3]. Music had ‘definitely’ been an important or meaningful part of the majority of care recipients’ (11/17) and caregivers’ (11/17) lives in the past. Music had been ‘somewhat’ important or meaningful for the remaining care recipients (6/17) and the remaining majority of caregivers (5/17), with one caregiver reporting music was not an important or meaningful part of their life in the past. Most participants reported no prior experience with music therapy. One dyad had previously completed a home-based music intervention in which a music therapist trained the caregiver to use caregiver-delivered music.

4.2 Procedure

After participants had enrolled into the study, we sent an online Plain Language Statement and Consent Form to caregivers, who confirmed whether the person they were caring for could sign their own consent form. If the care recipient was able, they could also read a plain language statement and sign the online consent form. Consent forms provided detailed information about the procedure of the full 8 weeks of the trial, including consent for recording interviews. Caregivers were subsequently sent a pre-trial assessment asking for basic demographic data, background questionnaires related to dementia diagnosis and severity, and information about their use of music.

We conducted semi-structured interviews with caregivers (and care recipients, if they chose) at Week 4 (the mid-point) and Week 8 (end of the trial). Although the app was designed to be used primarily by caregivers, the research focused on the care partnership and both care partners were invited to take part in all interviews. However, only one care recipient was present for the interviews and able to respond to questions. The interviews were conducted via Zoom and investigated how participants were using the app, what benefits they were experiencing, and what challenges they encountered that impacted ongoing use of the MATCH app. The interviews were guided by 13 questions, examining caregivers’ overall experiences with the app. We asked about caregivers’ daily routines, how they found the training content, what worked and did not work for them when using the app, what circumstances led to positive and negative experiences with MATCH, and if and how MATCH impacted interaction with the person they care for.

For the purposes of the present paper, we focused on the insights gained from these interviews, which were conducted at crucial points in the trial for understanding how participants were using the app and the impact it was having on participants’ care practices. We also collected other data to support the broader project aims of evaluating the app. These included: 1) a pre-trial interview conducted at the start of the trial, following the initial screening and assessments, which included a demonstration of the app conducted via Zoom; 2) a participant diary consisting of targeted questions on what worked and what did not work when using the app each day; 3) weekly contact with a music therapy research team member to check for any adverse events and to troubleshoot any problems with the app or use of music; and 4) post-trial assessments conducted following the final interview in Week 8. Participants were also invited to attend follow-up workshops to contribute to the design of the next version of MATCH. Due to the volume of the data collected and the focus of this paper, we have not reported on the data gathered through the pre- and post-trial assessments, participant diary, weekly check-ins, or follow-up workshops. These findings will be reported in future outputs.

4.3 Analysis

All interviews were video and audio recorded. Audio recordings were transcribed and checked by the first author. We analysed the transcripts using Braun and Clarke’s 6-phase approach to reflexive thematic analysis [8, 70]. After familiarisation with the transcripts (phase 1), the first author coded the data (phase 2). Codes were subsequently arranged into initial themes (phase 3) and discussed and reviewed with the rest of the research team (phase 4). Following multiple rounds of discussion, themes were further refined, defined, and named (phase 5). Further refinement occurred during writing, and final discussions with the research team (phase 6).

5.1 MATCH strategies benefited caregivers and people living with dementia

The MATCH application aims to support dementia care by teaching music therapy-informed strategies for using music in care. Caregivers described a range of benefits they experienced from using the strategies learned through the app. These included: 1) adding a new set of strategies to the caregiving toolbox; 2) enabling management of day-to-day care; and 3) enhancing the partnership between the caregiver and the person being cared for.

5.2 Making MATCH Work

The 8-week trial revealed the work required to ensure MATCH worked within participants’ everyday care routines. For some, this was quite successful – the app aligned well with existing care practices or technology setups. For others, this was more challenging. In this section we discuss themes that relate to how participants made MATCH work in their individual situations, in terms of: 1) integrating into existing contexts of care; 2) integrating into existing technology ecosystems; 3) different perspectives on creating playlists; and 4) how MATCH prompted reflections for future music use.

6.1 MATCH as a therapeutic tool

As intended, MATCH taught caregivers to use music in purposeful ways to address the needs of the person they cared for. Previous research in HCI has similarly shown that music and sound technologies can be integrated into the lives of people with dementia and support active engagement with music [33, 46, 48, 77]. Our study extends this prior work by demonstrating that strategies informed by music therapy principles can be particularly valuable for caregiving [4], and that family caregivers can benefit from technology-based training to understand these principles. Further, this research highlights the importance of the care partnership in trialling a new application for supporting dementia care. Key benefits that caregivers experienced from using MATCH strategies included supporting the partnership between the caregiver and the person they cared for, creating a more positive care environment, and supporting relationship-based care. Previous HCI research has shown that technologies can be designed to support dyadic interactions in care settings [50], and the potential for music technologies to promote connection within care partnerships has been identified in previous research [33, 48]. This speaks to an opportunity in HCI to ensure technologies for supporting care are accessible, flexible, and manageable for caregivers to use to support their caregiving [6].

Prior research has identified the effectiveness of providing informal family caregivers with training in the intentional use of music to support care, particularly with regards to managing agitation and other behavioural and psychological symptoms of dementia [1, 2, 4, 54]. Our study extends prior research on indirect music therapy by highlighting that technologies can be effectively designed and used to provide caregiver training, expanding participants’ strategic use of music for a range of care needs. Together, our findings support the use of MATCH in care, and highlight opportunities for arts-informed technologies to support caregivers and people living with dementia in terms of managing individual needs, as well as within the caring relationship [11].

6.2 How MATCH and music influenced caregiving

While MATCH was successful in achieving its primary aim - that is, teaching and providing caregivers with a new tool for day-to-day care - participants also described how making the app work was an ongoing process over the 8-week period. This included the process of implementing the music therapy-informed strategies explained in the training, and in creating the playlists suggested in the app. Our findings illustrate the importance of understanding the variety of individual and social contexts and situations in caregiving, and how these play a role in adoption of a new technology for care [36, 77, 81]. Furthermore, while HCI research has trialled a range of music-based technologies to support people living with dementia in residential aged care [31, 34, 46, 47, 48] it is evident that the home context presents its own unique challenges. Technologies that can be applied in one setting may not offer the same opportunities in the other.

Meeting the needs of people living with dementia in recognising personhood and agency around initiating music listening [82], was a key concern for some participants. Ensuring choice in how both the strategies and technologies were used was empowering for some of the participants living with dementia, echoing previous work [35, 77]; however, this was a challenge for some caregivers if their loved one did not always remember how to put music on. While MATCH was primarily designed to be used by caregivers, Nina’s example, where her mother was agitated in response to waking up to music she had not turned on herself, illustrates the importance of recognising personhood when designing and trialling technologies to be used with people with dementia [41, 79]. Recognising personhood has to be balanced with playing music for those who may struggle to initiate activities [82]. Caregivers described other ways of including care recipients, discussing how using the app could facilitate new conversations and activities with their loved one, consistent with existing HCI research where technologies can be used to stimulate social agency [26].

Considering the daily use of the MATCH app, caregivers described how the app did and did not fit into their existing music technology ecosystem. Previous research has identified a range of formats and devices that older adults and aged-care residents without a dementia diagnosis use for music listening [39], and we found that caregivers and people living with dementia similarly had diverse preferences around music technologies. Trialling MATCH altered some participants’ preferences, yet for others, their preferred music technology influenced whether they used the app in an ongoing way. These personal preferences impacted how music and MATCH strategies were used, illustrating the importance of considering the varying technologies participants use to play music when introducing a new application. Understanding how new digital tools fit within intended users’ existing technological ecosystems is essential for ensuring new therapeutic digital tools are successfully adopted [36, 81].

Alongside consideration of the music technologies used by participants, we encountered interesting findings around how participants created and used playlists. MATCH aimed to support dementia care using music therapy-based principles, and these therapeutic ideas were brought into the app design and intended usage. We assumed that the suggested playlists and the training encouraging creation of personalised playlists, would give caregivers ideas for building their playlists. Diverse experiences and usage were evident, with some caregivers using a mix of the suggested playlists and their own personalised playlists, and many appropriating module playlists to suit their own purposes. This contrasts with research that suggests benefits from personally meaningful music for people living with dementia, from their adolescence (i.e. the ‘reminiscence bump’) [40, 58]. Previous research on training systems to support people living with dementia has similarly found unexpected observations and appropriation of such technologies [76]. This suggests that despite participants having guidance and research to support the intended outcome, practical constraints may lead to different usage in practice. Our findings also illustrate that the values and assumptions that guide design of technology-based therapeutic tools are not always shared by participants [81].

Using MATCH inspired caregivers to reflect on their music listening habits, and how these might change in the future. Some caregivers even began to consider how they consumed digital media broadly, applying the strategies beyond music listening. These reflections highlight a range of further contextual and technological factors that must be considered when implementing new digital tools into care, with participants discussing plans for integrating the app into different times of day, and adopting additional technologies to expand their music listening. This suggests that MATCH, and other music-based technologies, need to be easily adaptable and compatible with a broad range of devices, to support different preferences for music technologies. This finding, alongside questions from participants on the ongoing use of the app, aligns with research suggesting that technologies can be designed to perform a ‘gateway’ effect [64]. That is, a new technology may be valuable to initiate new behaviours, but may not be needed longer-term. For some heath technologies, abandonment could be a marker of success, suggesting that the therapy or intended behaviour change, has been completed [15, 65]. While MATCH was intended to encourage ongoing use, with content that should apply to a range of situations and changes associated with dementia, these findings raise interesting questions around how the app might be used after completing the module content. If the app is not always used by caregivers to play recorded music, this does not necessarily mean the strategies embedded in the app have been abandoned.

6.3 Insights for Designing Therapeutic Music Technologies for Dementia Care

Drawing on our findings and previous literature, we identify four lessons for designing music-based digital tools to support the care of people living with dementia. Each of the four lessons aligns with the overarching insight that personalisation is crucial when designing technologies for dementia care. As previous researchers have noted, there is no “one-size-fits-all” solution when it comes to designing and using technology in dementia care [30, 61]. Our study suggests this is further complicated when designing music technology for use in dementia care, particularly when the underlying aim is to use music therapeutically to support the needs of both caregivers and care recipients. We argue that music technologies for dementia care need to be designed to accommodate diversity in: the care partnership, music and technology preferences, therapeutic goals, and dementia progression.

Designing formusic use inthe care partnership. First, our findings emphasise the importance of personalisation for the care partnership. Caring relationships, situations, and music preferences differ, and our study showed that these differences influenced how the app was used. New digital tools for music-based care therefore need to be adaptable to accommodate these differences. In our study, MATCH strategies supported the partnership between the caregiver and the person they cared for, and created a more positive care environment, especially when music was used to meet the needs of both caregivers and the person they were caring for. However, we also found instances when music preferences differed among members of the care network. Prior work suggests that technologies used in dementia care should be co-designed with a range of stakeholders in the care network alongside people living with dementia. [21, 50, 84]. Our findings build on this work, highlighting that in the case of a music therapy-informed technology for dementia care, we must consider the care partnership, and the broader care network, rather than focusing on either care recipients or caregivers as the primary users. This is challenging when designing music technologies, as music preferences differ; therefore, future work needs to consider how music technologies can be designed to accommodate different preferences and needs within the care partnership.

Designing for individual preferences in music technology use. Second, participants expressed diverse personal preferences for both music and technologies, which influenced how caregivers trialled music and MATCH strategies. Some caregivers fully embraced the training and strategies, while others were more ambivalent, selecting relevant strategies and methods of playing music to fit within their home and technological contexts. These individual differences in the use of the app demonstrate that new digital tools must be adaptable and able to be personalised to a range of individual and household contexts - music technologies informed by therapeutic goals cannot be ‘one-size-fits-all’ [30, 61]. This has implications for further iterations of music therapy-informed technologies such as MATCH, to ensure that caregivers and people living with dementia can use such technologies within their existing personalised ecosystems, and on a range of devices including smart speakers and TVs. Further, digital music tools must be adaptable to suit a range of music preferences — for example, by not limiting recommendations to music from the reminiscence bump. These tools must be able to be personalised, and easily appropriated as needed to support changing goals and needs associated with dementia [25].

Designing for different therapeutic goals. Third, we found that participants’ use of the MATCH app sometimes differed from the therapeutic assumptions implicit in the design, because participants had their own goals. Unlike other dementia care technologies examined in HCI, MATCH was designed alongside music therapists who had clear views about the therapeutic and training goals of the app. However, time constraints and individual preferences of participants led to differences in use of the app, particularly around building playlists. Previous research has discussed the importance of participatory design to account for such individual differences and interests [49], and how these may influence users’ approach to a technology [7, 52]. Building on prior work and our current findings, future research will continue to iterate the design of MATCH. To account for these personal differences, we need to provide a greater number of well-chosen suggested songs for busy caregivers who favour ready-made playlists, and allow for the autonomy that others prefer. Furthermore, despite the training encouraging not just music listening but other activities such as singing, some participants did not use these strategies. Instead, they gravitated towards using playlists of recorded music, even when setup sometimes took too much time, and strategies such as singing would be more accessible when dealing with care crises. The format of MATCH as a mobile application may have influenced this. Singing spontaneously to calm a difficult situation does not require the caregiver to use an app, and participants may have overlooked this strategy because they were trying to “use” the app. Even when technologies are designed to align with therapeutic principles, the principles can be practised in different ways, which may be influenced by the format of the technology. In the case of MATCH, future content may aim to address this, with additional instruction within the training on the music therapy principles guiding the strategies, and future co-design considering suggestions around expanding functionality to allow for greater flexibility in using music for a broader range of therapeutic goals.

Designing for different needs across the progression of dementia. Fourth, our findings emphasise the importance of personalising music therapy-informed tools for the dementia stage of the care recipient. We found that for some care recipients, it was important for them to be able to choose their own music, and when music was played. While not all caregivers experienced problems where the person they cared for was agitated by music (as in the case of Nina and her mother), many still found benefits from ensuring their family member was involved in choosing music for themselves. This emphasises the importance of recognising personhood when designing and trialling technologies to be used with people with dementia [25, 41, 79], and carefully balancing this with supporting those who might need assistance with initiating music listening [82]. We were limited in the present study by being unable to investigate care recipients’ experiences with MATCH, thus future research must consider ways to ensure input from participants living with dementia, regardless of dementia stage [25]. Additionally, given that there is growing interest in the use of artificial intelligence for supporting dementia care [53], it is important that the need to maintain autonomy and recognise personhood in people with dementia is supported by future smart music technologies. For instance, if a system is designed to automatically play music in response to detected signs of agitation or distress for someone with dementia, this may have benefits for some people, but negative effects for others. It is crucial that researchers and technology developers are mindful of the need to maintain and support care recipient autonomy wherever possible.