Abstract
Intelligent assistive technologies (IATs) have the potential of offering innovative solutions to mitigate the global burden of dementia and provide new tools for dementia care. While technological opportunities multiply rapidly, clinical applications are rare as the technological potential of IATs remains inadequately translated into dementia care. In this article, the authors present the results of a systematic review and the resulting comprehensive technology index of IATs with application in dementia care. Computer science, engineering, and medical databases were extensively searched and the retrieved items were systematically reviewed. For each IAT, the authors examined their technological type, application, target population, model of development, and evidence of clinical validation. The findings reveal that the IAT spectrum is expanding rapidly in volume and variety over time, and encompasses intelligent systems supporting various assistive tasks and clinical uses. At the same time, the results confirm the persistence of structural limitations to successful adoption including partial lack of clinical validation and insufficient focus on patients’ needs. This index is designed to orient clinicians and relevant stakeholders involved in the implementation and management of dementia care across the current capabilities, applications, and limitations of IATs and to facilitate the translation of medical engineering research into clinical practice. In addition, a discussion of the major methodological challenges and policy implications for the successful and ethically responsible implementation of IAT into dementia care is provided.
Keywords
INTRODUCTION
The increasing prevalence of dementia poses a major challenge for global health at multiple levels. At the financial level, Hurd and colleagues (2013) calculated that dementia and specifically AD are among the most expensive diseases for Western society, with a price tag per year of around $160 billion [1, 2]. As the greatest relative cost increases are occurring in low-income African and in East Asia regions [3], the provision of dementia care services will be seriously exposed to danger due to pre-existing limitations of local national budgets.
Long-term care at nursing homes and other healthcare institutions is a major component of this societal and economic burden whose impact affects not only public finances but also the provision of healthcare services. In addition to institutional care, a large proportion of dementia care is provided by informal caregivers (usually family members). In the U.S., more than 15 million Americans provide unpaid care for family members affected by AD and other types of dementia [4]. At the individual level, these informal caregivers often experience psychological burden, with more than 40% of them reporting emotional stress and 74% reporting concern about maintaining their own health since becoming a caregiver [4]. In 2014, this unpaid informal contribution provided an estimated 17.9 billion hours of medical and social assistance, hence providing an overall value nearly equal to the costs of direct U.S. medical and long-term care of dementia [1]. However, this crucial component of dementia care provision is expected to rapidly shrink as a consequence of demographic trends. Currently, the potential support ratio (PSR)— defined as the number of people aged 20-64 divided by the number of people aged 65 and over— is under 4 in most North American and European countries [3]. By 2050, at least 35 countries will have PSRs below 2, hence having fewer than two people under 65 for each senior person. In the context of dementia, the caregiver-to-patient ratio is expected to reduce accordingly [5]. This progressive scarcity of human caregivers will put additional financial and logistic pressures on the healthcare systems. Finally, due to the highly disabling condition of their disease and the increasing limitations to the provision of care, the growing population of older adults with dementia and their caregivers will face major challenges to quality of life [2].
In response to this emerging global scenario, technological innovation is likely to be a critical factor. Recent advancements in artificial intelligence (AI), pervasive and ubiquitous computing (PUC), robotics, and mobile computing combined with new developments in wireless networking and human-computer interaction open the prospects of reshaping dementia care with intelligent technology. In fact, the pervasive deployment of intelligent assistive technologies (IATs) for dementia may have a disruptive impact on dementia care [6]. IATs could (i) mitigate the burden on public finances through the delay or obviation of institutional care, (ii) reduce the psychological burden on formal and informal caregivers, (iii) compensate for the progressive scarcity of human caregivers while enhancing and optimizing quality of care, and (iv) empower older adults with dementia and thereby improving their quality of life.
Assistive technology is the umbrella term used to describe devices or systems which allow to “increase, maintain or improve capabilities of individuals with cognitive, physical or communication disabilities” Marshall [7]. IATs are assistive technologies with own computation capability and the ability to communicate information through a network. Most of them display the ability to sense the external environment or digital ecosystem and provide adaptive responses in a manner that maximizes the benefits for the users (e.g., increasing safety). IAT encompasses a wide spectrum of technological applications currently used or in-development with potential application to dementia care. These include self-contained devices (e.g., tablets, wearables, personal care robots, etc.) and distributed systems (e.g., smart homes, integrated sensor systems, mobile platforms, etc.), as well as software applications (e.g., mobile or web-based apps). While AI provides systems capable to simulate aspects of human intelligence, PUC embeds intelligent microsystems into everyday objects and homes whose friction with the user is progressively mitigated by advancements in human-computer interaction. In parallel, intelligent service robots can assist users in a variety of dimensions including personal care, companionship, and social and emotional support.
Although IATs open up promising prospects for the future of dementia care, their adoption is still lower than expected [8]. This has been attributed to suboptimal information transfer and dissemination across technology development and medical implementation [9] as well as to the lack of solid and highly generalizable clinical validation of many IATs [10]. Clinicians and other health professionals are often unaware of new IATs and their applicability to dementia care, as little cooperation has occurred between technology development and medical implementation. In addition, a mismatch between the user’s cognitive profile and the prescribed IATs has been reported as a consequence of top-down approaches to technology design [11].
Concomitantly, the prevalence of participatory and user-centered (UC) approaches to technology design is reportedly low [8]. The “user-centered” or “patient-centered” approach is a framework for the design and development of new products or for the assessment and evaluation of existing products in which the needs, wishes, and limitations of end-users of the IAT are given extensive attention at each stage of the design or assessment process [12, 13]. Such low prevalence of participatory and UC approaches to technology design has been observed as a co-determinant of low adoption rates since it obstacles the incorporation of end-users’ needs, desires and wishes into product development [14]. Finally, clinical studies designed to validate IATs are often affected by structural limitations and methodological weaknesses including small sample-size, high drop-out rates, low statistical significance, and inadequate adjustment for multiple comparisons [10]. This lack of solid and highly generalizable clinical validation might contribute to the slow translation of emerging IATs from the designing labs into the clinics.
While the number of publications in this area has been significantly growing in the past few years [15], yet a comprehensive and up-to-date index of IATs with possible application into dementia care has not been produced. Given the pace of innovation in medical technology and the reported translational gap between bench and bedside, the production of a comprehensive index is highly needed to orient health professionals, affected individuals and other operators involved in the provision of dementia care across this rapidly emerging domain. Such a state-of-the art index will provide comprehensive information for relevant stakeholders about current IAT possibilities and limitations hence contributing in raising awareness about the pros and cons of their integration into care. While one study recently reviewed dementia-focused assistive technologies [16], their review did not result in the production of a technology index. A cognitive function based review of assistive technologies for cognitive impairment retrieved only 13 studies with direct focus on assistive technologies for dementia, hence covering a very small portion of the currently estimated technological spectrum [17]. In addition, both studies failed to distinguish IAT from other assistive technology (e.g., tools without computing capacity), thus underestimating the specificity of intelligent technology and its potentially game-changing role in the clinical setting. Other previous reviews have either not been systematic [18] or have limited their scope to a subsection of assistive technology for dementia such as “assistive devices for the hours of darkness” [19]. The latest comprehensive list of IATs (especially cognitive orthotics and advanced integrated sensors) with potential applications to dementia care dates seven years back [20]. As technology evolves fast, a new up-to-date index is urgently needed to keep up with advancinginnovation, guide health professionals across emerging technological opportunities, and contribute towards adequate uptake of potentially useful tools to improve the lives of older adults with dementia.
In order to facilitate the successful implementation of emerging intelligent assistive technologies into dementia care, such index should address six critical questions: (i) How large is the current IAT spectrum and what is its growth rate over time?; (ii) What types of device are currently available?; (iii) What cognitive or physical functions can be assisted through IAT?; (iv) Which patient population segments are targeted as end-user groups?; (v) Which is the current level of clinical validation of existing IATs?; (vi) What approach is prevalent in the design and development of such devices?
In the following, we address these critical questions by presenting the results of our systematic review and aggregate such information into a comprehensive technology index. This index is designed to provide comprehensive information to clinicians, researchers, patients, caregivers, and other stakeholders involved in dementia care about the current possibilities and limitations of intelligent assistive technology for dementia.
METHODS
Data search and extraction
A literature search was performed for English language articles indexed in the following search engines and bibliographic databases: IEEE, PubMed, Scopus, PsycINFO, and Web of Science. We searched title, abstract, and keywords for the terms: (“assistive technolog*” OR “assistive device” OR “assistive application”) AND (“intelligent” OR “ICT” OR “adaptive” OR “computer” OR “robotic”) AND (“Alzheimer*” OR “dementia” OR “ag*ing” OR “elder*). Query logic was modified to adapt to the language used by each engine or database.
Studies included in the synthesis had the following features: (i) original articles, book chapters or conference proceedings; (ii) written in English; and (iii) published between January 1, 2000 and April 12, 2016. Additionally, studies included in the synthesis must present the (a) design and development, or (b) assessment and evaluation of one or more intelligent assistive systems with current or potential applications to dementia care. Reviews, commentaries, letters to the editors, and opinion articles were removed.
Intelligent assistive systems included into the technology index met the following inclusion criteria: (i) had their own computing capability 1 ; (ii) had direct applicability to dementia care; and (iii) could assist or compensate for one or more functional impairments associated with AD or other age-related dementias (e.g., memory loss and executive dysfunction). Reviewers excluded non-intelligent systems (i.e., without own computational capacity, such as walking canes or printed cognitive training books) as well as systems developed for the support of non-progressive traumatic brain injuries with limited applicability to the cognitive, emotional, and physical deficits specific of AD and other age-relateddementias.
A total of 617 papers were initially identified. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) criteria [21], four steps of filtering were performed (See Fig. 1). First, additional 167 records were identified by reviewing the references of all initially retrieved articles. Second, duplicates were removed using the ENDNOTE tool for duplicate detection. Additional duplicates were removed manually after reviewing the abstracts. A total of 5 duplicates were detected. Second, eligibility assessment was performed on the remaining 779 papers to remove sources that did not meet the review’s inclusion criteria. Further 208 publications were rejected at this stage of filtering. Third, in-depth review was performed on the full-text articles of the remaining 571 entries included in the synthesis. Fourth, articles were clustered according to the categorization criteria described below. Two reviewers performed all stages of filtering independently, and only the papers rejected by both reviewers were removed from the working corpus.
Categorization
In order to produce a systematic technology index, retrieved technologies were clustered according to the following main characteristics: (i) technology type; (ii) application; (iii) function assisted; (iv) user-centered design; (v) primary target-user population; and (vi) evidence of clinical validation. For already marketed products the official commercial names were adopted. In contrast, prototypes without commercial name were listed according to the names or descriptions that the study authors used.
Technological types categorize IATs according to their hardware or software architecture, composition (single device versus distributed system), and type of human-machine interaction (e.g., wearables versus hand-held devices). We grouped them into seven major categories: distributed systems, robots, mobility and rehabilitation aids, hand-held multimedia devices, wearables, human-machine interfaces, and software applications. A distributed system is defined as a system composed of several sensing and processing sub-systems, which communicate through a computer network, “hosting processes that use a common set of protocols to assist the coherent execution of distributed activities” [22]. This category includes ambient assisted living (AAL) systems, i.e., distributed assistive systems based on the ambient intelligence paradigm that shape the user’s environment in a manner that is sensitive, adaptive, and responsive to their needs. When AAL systems are used to reshape and augment the user’s home (e.g., with sensors, wireless networks, and software applications for healthcare monitoring), this is called a smart home. Personal robots are autonomous service robots designed for the use and benefit of individuals. These include domestic service robots that support Activities of Daily Living (ADL), robotic cognitive assistants, and companionship and socially assistive robots that support, respectively, the relational and social dimension of patients. While robots are autonomous individual agents, mobility aids are mobile machines (e.g., powered wheelchairs) or worn assistive systems (e.g., exoskeletons) that can facilitate mobility, limb movement, and control for users with physical disability. Hand-held multimedia devices are mobile devices having a display screen with touch input and/or a miniature keyboard, hence usable with hand control. These include smartphones, personal digital assistants (PDAs), tablets, and other multimedia technologies capable of generating text, audio, images, animation, video, and interactive content. Finally, wearable devices are technologies incorporated into items of clothing and accessories worn by the user, hence represent the most intimate and closed-up form of non-invasive human-computer interaction.
Applications categorize the primary type of activity supported by the system and its primary functionality. In contrast, the category “function assisted” determines the specific functional impairment caused by or associated with AD or other dementias to which the IAT provides compensation or assistance. When IATs were designed for the compensation of more than one single cognitive, emotional, or physical deficit, they were categorized as general-purpose systems.
Finally, the target user category indexes IATs according to the specificity of the primary end-user population targeted by researchers when designing the system. In fact, although all IATs included in the index had direct application as compensatory tools for one or more functional impairments associated with AD or other age-related dementias, some of them may not be designed exclusively for AD patients but with a more inclusive population target.
In addition, the reviewers determined for each IAT whether it was developed and designed following a user-centered approach and including participatory design techniques. Since UC design has often been considered a major predictor of social adoption and a crucial factor of ethically sustainable technology development [14, 23], we investigated what proportion of the entire spectrum of IATs for dementia incorporates such approaches. A logistic regression was performed on the correlation between time and the frequency of user-centered models of technology design with the purpose of testing if the adoption of user-centered models is increasing over time. Finally, following several reports on the lack of evidence-based studies on the clinical effectiveness of IATs for dementia [15, 24], the reviewers investigated if each IAT had been preliminary validated via clinical studies on human subjects.
RESULTS
Our review identified 539 IATs with current or potential applications to dementia care. These systems are included into the technology index summarized in Table 1 and analyzed according to the previously listed core categories. For each IAT, their technological type, function, assisted deficit, a user-centered design, primary end-user population, and evidence of clinical validation are presented.
Expansion of the IAT spectrum (2000-2016)
Results show the number of IATs is rapidly increasing over time, hence confirming the progressive expansion of the IAT trend in dementia care. As represented in Fig. 2, the number of IATs with application to dementia care has increased by over 6 times in the period 2006-2010 as compared to 2000-2005 and has even increased by a factor of 15 in the period 2011-2015 (since the literature review was carried until April 12, 2016, only approximately one-third of the IATs published in 2016 are captured in Fig. 2).
Of the total, the majority of systems were distributed systems (n = 194), followed by robots (n = 97), and mobility and rehabilitation aids (n = 62). System goals were determined based on their primary capability according to the intent of the manufacturer. Most IATs were designed with the purposes of supporting users in the completion of ADLs (n = 148), monitoring users and their environment (n = 100), or providing, respectively, physical (n = 88) and cognitive assistance (n = 85). Assisted deficits were determined according to the primary cognitive or physical deficit associated with dementia to which the IAT provides compensation. Of the total, most devices were general-purpose, i.e., non-modular but broadly applicable across psychophysical domains, without specialized features exclusive for a particular domain (n = 250). Devices exclusively programmed for specific deficits include motor function (n = 109), impaired cognition (n = 140), and mood and emotional disturbances (n = 31). The category “end-user population” was determined according to the end-user segment explicitly targeted by the researchers. Our review identified four end user types according to the specificity of the target population chosen by the researchers: general elderly and disabled population, people with dementia, AD, and specific stages of AD. Models of design and development or assessment and evaluation were screened to identify the proportion of IAT adopting user-centered design and assessed through clinical studies. The variation over time of UC approaches was also examined. Results reveal that the prevalence of UC approaches to IAT design and development is significantly increasing over time and that half of the IAT spectrum received preliminary clinical validation. In the following, we present these findings in detail.
Technological type
With 194 items, distributed systems represent the largest proportion of IATs for dementia (Fig. 3). These include smart service platforms and AAL technologies, i.e., distributed assistant systems “for the constitution of intelligent environments” that “aim to compensate predominantly age-related functional limitations of different target groups” [25]. An example of AAL distributed system is Zhang and colleagues’ (2014) Smart Assistive Living (SAL) platform. This platform is designed to support the delivery of telehealth and telecare services to older people suffering from dementia, and can enable them to stay at home longer and more independently [26]. The digitalization of the domestic and residential environment is also accelerated by the application to healthcare of the Internet of Things technology, another important subcategory of distributed systems (n = 24).
With rapid advancements in medical robotics, personal care robots represent the second most common technological type (n = 97). While most reviewed robots are domestic service robots (n = 62), i.e., autonomous systems that assist users in the completion of practical activities such as house maintenance, alarming, telehealth, etc., an interesting growing portion of robots (n = 31) is being designed to assist the emotional and social dimension of older people with dementia. Robots of these type are called socially assistive robots [23, 27]. A successful example is robot PARO, developed by AIST (See: http://www.parorobots.com/; last accessed: June 2, 2016). Designed to stimulate patients with AD and other cognitive disorders by providing emotional assistance and companionship, PARO has been effectively applied as part of standard occupational therapy [28] and revealed a positive effect on the residents’ quality-of-life and pleasure scores [29]. All reviewed robots including PARO presented some degrees of AI such as the ability to learn and remember their own name, and to learn when their behavior results in positive responses of the user.
The relative frequency of hand-held multimedia devices (n = 50) is presumably facilitated by the availability of these devices among the general population as everyday communication tools, as a consequence of the growing importance of such mobile devices in our societies. These technological types: include smartphones, tablets, PDAs, and other mobile devices, co-evolving with a rapidly growing digital ecosystem of compliant mobile apps and other software applications (n = 50). At the software level, in fact, most hand-held devices were designed to run specific assistive mobile or web applications programmed for people with dementia. For example, the SmartBrains mobile app was developed to provide cognitive enhancement for people with AD [30] and was reported to “greatly augment” the “traditional psychomotor stimulation” [31].
Wearable devices (n = 44) accounted for a slightly smaller portion of the IAT spectrum, possibly as a consequence of the most recent growth of this technological trend. However, since the number of wearables is increasing rapidly over time, it is reasonable to predict that such applications will play an increasingly prevalent role in technology-assisted dementia care. One promising application is the incorporation of a wrist wearable unit into an Android smartwatch to monitor the physical activity of the user and enhance independent living [32]. Neurowearable devices such as virtual reality (VR) and augmented reality (AR) systems composed a smaller proportion of the IAT spectrum.
Finally, Human-Machine Interfaces broadly encompass the realm of hardware and software systems (n = 42) designed to establish a direct connection pathway between the human user and an external computer device. Among those, particular interest from the perspective of neurocognitive rehabilitation is raised by brain-computer interfaces (BCIs), as they allow users to control external devices solely with brain activity (usually via EEG-recordings), hence bypassing the peripheral nervous and muscle system [33]. Since BCI applications are usually based on instrumental learning and require users to self-regulate their brain activation— a task whose completion is very limited among elders with dementia, they were considered for many years not suitable for compensating the cognitive deficits in AD patients [34]. However, recent advancements in the detection of involuntary brain signals (e.g., related to emotional states) are enabling the development of new BCI solutions (n = 7) with possible application to dementia care, including BCIs for early diagnosis, computerized cognitive training and communication [34, 35].
Application
IATs are currently being implemented into dementia care for a variety of purposes (Fig. 4). Our review results reveal that the most common application of IATs in dementia care is supporting older adults with dementia in the completion of ADLs (n = 148) such as eating, bathing, dressing, toileting, and continence. These results reflect the oft-stated wish of elders with dementia to enhance their independent living and the need for healthcare systems to delay or obviate institutional care, hence age-in-place [36]. With 100 systems, monitoring is the second most common application. Monitoring is a key function for enhancing a person’s safety as it allows identifying patterns of abnormal behavior, prompting responses from caregivers in case of danger and collecting data for other connected applications. Physical (n = 88) and cognitive assistance (n = 85) also compose an important proportion of the overall applications of IAT to dementia care. Cognitive assistants are intelligent devices capable of supporting or augmenting cognitive functions in cognitively impaired individuals, functioning as external cognitive processors. These include memory aids and other cognitive orthotics. An example is the COGKNOW Day Navigator, a digital prosthetics to support persons with mild dementia in their daily lives, with memory, social contacts, daily activities, and safety [37]. In contrast, physical assistants compensate for motor and locomotive deficits associated with dementia-related disability. An example is the MOBOT, an intelligent physical assistant to support elderly patients with mobility disabilities during gait and sit-to-stand transfer [38]. Emotional support and assistance represents a smaller (n = 15) but rapidly developing portion of IAT application. Finally, promoting interaction (n = 64) and engagement (n = 22) as well as facilitating care and rehabilitation complete the picture of possible applications enable by current IATs for dementia.
Function assisted
Most reviewed systems appeared to be general-purpose (n = 250), showing how the complex disabling condition of dementia affects in parallel various components of a person’s psycho-physical dimension (Fig. 5). With 140 items, the cognitive dimension (encompassing not exclusively executive function but also perception and communication) is the component of dementia-induced disability most commonly assisted via IAT. Among these cognitive faculties, memory predictably scores first (n = 33), followed by communication (n = 28), orientation (n = 18), reasoning (n = 12), and decision-making (n = 8). Physical assistance such as assistance in mobility, navigation, and motor control represents the third most common category of deficits assisted by IATs. Dementia-associated disturbances of the emotional and affective sphere are supported by a significant portion of IATs (n = 31), revealing an increasing effort to compensate through intelligent technology an often-neglected component of care provision for older adults living with dementia. Finally, a significantly smaller number of IATs (n = 9) can provide assistance to the social dimension of elderly adults with dementia by reducing isolation and facilitating social interaction.
User-centered design
Results show that, to date, only 40.1% of reviewed IATs are explicitly designed, developed, or assessed through UC approaches (see Fig. 6). Among those, cooperative design and participatory design approaches are often recognizable, with researchers and users being involved on an equal footing in the various stages of the design process [39]. However, the results of our logistic regression show a statistically significant correlation between time and the frequency of user-centered models of technology design and assessment (b = 0.21, Wald(1) = 6.17, p < 0.01). Therefore, this correlation predicts that the prevalence of UC approaches will become majoritarian in IAT design and development in the near future.
Primary target population
As different forms of dementia and different stages of the disease progression may present inherently specific symptoms and care requirements, we looked at the level of selectivity adopted by the technology designers in determining the end-user population of each IAT (Fig. 7). While all reviewed IATs could assist or compensate for one or more functional impairments associated with AD or other dementias, most of them were not exclusively designed for people with AD or other forms of dementia but also for the general elderly population with neurocognitive disability (n = 362). A significantly smaller portion (n = 115) of intelligent systems was more selectively designed to primarily target people with dementia, or the specific cognitive, physical, and behavioral symptoms of people with AD (n = 51). Finally, only nine devices were specifically designed to target either specific stages of AD (n = 5) or mild cognitive impairment (n = 5).
Clinical validation
Our analysis of the clinical validation of IATs for dementia shows that little more than half of them (50.65%) did not receive clinical validation through clinical trials with human subjects. Among the subset IATs that received clinical validation (n = 266; 49.35%), most validation studies (n = 254) were conducted with small sample sizes (< 20 participants). Randomized-controlled design was reported in only 1.1% of clinical trials, that is 3 studies.
DISCUSSION
In comparison to previous reviews [8, 17], our results show that the spectrum of IATs for dementia is expanding fast in volume, variety, and potential applications. Since the number of IAT applications is approximately doubling every five years, IATs are likely, in the near future, to become a ubiquitous trend in dementia care. In addition, as the overall processing power for computers is increasing linearly over time and rapid advancements in micro-computing are accelerating the miniaturization of IAT systems [40], the expansion of the IAT spectrum will be accompanied by a coordinated performance potentiation. Such computing trends will generate novel possibilities for people suffering from dementia to live more independently, autonomously, and safely, and will facilitate the delivery of care to this growing patient population.
At the level of product development, the large proportion of AAL technology and other distributed systems attests an ongoing smart-environment trend in healthcare. As our results report, pervasive and ubiquitous computing techniques are being increasingly used to incorporate automation into domestic and residential environments with the purpose of delivering services, improving efficiency, performing or facilitating daily activities, and improving the wellbeing of their residents. After the Internet of Things has already incorporated computer technology into everyday objects such as televisions and other electronic appliances [41], the transition to smart homes could represent the next disruptive change in domestic environment. Although these trends are not exclusive to healthcare but common to various aspects of modern societies, i.e., through the creation of smart cities or cybervilles, their application to dementia care is particularly promising since it could delay the need for long-term care and institutionalization, hence result in significant cost-reduction for healthcare finances and improved quality of life for the senior population [6, 8]. In fact, such distributed assistive systems have the potential to prolong the safety and independence of older adults with dementia, preventing accidents, assisting them during the completion of ADLs, facilitating caregiver supervision, and triggering alarm in case of emergencies [42].
Hand-held multimedia devices are also likely to play an important role in the future of technology-assisted dementia care. In the light of their widespread use as everyday technologies in modern society, smartphones and other hand-held devices are often recognized by users as familiar tools, therefore requiring a lower level of training— especially among baby-boomers, who are often smartphone savvy [43]. This is likely to result in higher social adoption, in particular if combined with coordinated advances in mobile software technology. In parallel, the recently increasing frequency of wearable devices underscores the need for a more widespread distribution of friction-free, non-invasive tools since non-obtrusive use is a reported priority among elders [44, 45]. Although the concept of wearable device dates back to the 80 s, only recent progress in miniaturization, micro-computing, and reduction of form factors allowed a significant development of wearables for commercial and medical purposes. However, while commercial applications of wearable technology are becoming increasingly popular among general users, the assistive and medical application of this technological type has not reached yet a sufficient level of maturity. In the next few years, a new generation of low-cost, friction-free, and information secure wearable devices is expected to add further opportunities to the current IAT spectrum. In contrast, assistive robots seem to have already reached a relatively high degree of commercial maturity, with several assistive robots— such as PARO (Daiwa House Industry), NAO (Aldebaran Robotics), Pepper (SoftBank), and PALRO (FUJISOFT)— already being commercially available in Japan, Europe, and the United States. Future advances in robotics, especially the use of adaptive intelligence and the problem-resolution at the human-machine interface, are likely to increase even more the distribution of these IAT types.
The high distribution of IATs for facilitating human-to-human interaction and human caregiving shows that some reported worries about the potential risk of “dehumanizing care” with assistive technology [46] are rather unjustified. In fact, technology-enabled care is not alternative to human-delivered care but complementary. For example, telepresence robots such as Giraff (Giraff Technologies AB), allow caregivers to virtually enter the home of a person with dementia from their computer via the Internet, hence supervising, monitoring, communicating, and conveying their presence as if they were physically there.
With the rapid erosion of the caregiver-to-patient ratio, the proportion of intelligent assistance to be integrated into general care is bound to expand significantly. This will require not only advancing the development of IATs for cognitive and physical assistance but for emotional support as well. As our results show, intelligent emotional assistants represent a minor proportion of IATs currently developed for compensating psychophysical deficits associated with dementia. However, developments in artificial emotional intelligence could dramatically accelerate the successful integration of IATs into standard care. As people with dementia often present emotional disturbances such as anxiety, depression, agitation, and distress [47], IATs programmed to learn “when and how to display emotion in ways that enable the machine to appear empathetic or otherwise emotionally intelligent” will be crucial for the future ofcare [48].
As the list of current applications shows, IATs are not only increasing in number but also in variety. While the first generation of IATs was primarily focused on promoting safety through tracking, alarm prompting, and remote monitoring (e.g., fall detectors and GPS trackers), current IAT applications are designed to support a number of activities including communication, telecare, and entertainment. In addition, the high number of applications for supporting ADLs shows that the main focus of most current IATs is not simply monitoring older adults with dementia, but empowering them by promoting the autonomous and successful completion of daily activities and the support of their psychosocial dimension (e.g., entertainment, engagement, and communication). Since emotional and psychosocial factors are recognized as important to stabilize mental health [49], this emerging holistic trend in IAT has the potential to achieve greater outcomes than earlier trends in technology-assisted dementia care.
From the perspective of the specific dementia-related deficits compensated for or assisted by IATs, the large prevalence of general-purpose systems (46.3% of the total) attests the complex and multifaceted condition experienced by elderly people with dementia. Since the disabling condition of dementia encompasses various components of a person’s physical, cognitive, and behavioral dimension, IAT solutions are often required to provide a holistic and multi-level support to their users. Therefore, advances in adaptive intelligence and other trends in AI are expected to be of extreme benefit for dementia care.
When defining the product’s end-user population, there is a need for a more narrow focus on the specific needs of each category of end-users. As our results show, many IATs tend to be targeting a vast and clinically heterogeneous end-user population including people with various forms of dementia and general neurocognitive disability. This fact may reflect the commercial advantage for producers to maximize the number of possible end-users for each marketed product. In contrast, the proportion of devices selectively designed to support specific stages of dementia is currently low. This lack of specificity could represent a significant obstacle toward the massive adoption of IATs for dementia and could add an additional reason to the limited uptake of IATs. In fact, patients suffering from different stages of AD may present specific needs and limitations that may be qualitatively and/or quantitatively different than those of people with other age-related cognitive disturbances or disabilities [50]. Also, people with mild AD often present different needs and limitations than people with moderate or advanced dementia and vice versa. Future IATs should be adaptive to each specific form of dementia and, within the same form of dementia, to each specific stage of progression of the disease in order to better reflect the specific needs of each end-user group and sub-group.
Another limitation of current IATs emerging from the review is the scarcity of clinical trials assessing the clinical effectiveness and safety of each product. As every second IAT for dementia lacks clinical validation, health professionals and institutions may be reasonably reluctant to introduce IATs into standard care. In addition, among the subset of IATs with reported preliminary validation, several clinical studies reported major limitations in terms of sample-size, drop-out rates, statistical significance and adjustment for multiple comparisons. While technical feasibility and usability were successfully tested via simulations in approximately all reviewed IATs, well-designed, statistically-significant and highly generalizable randomized controlled studies with older adults with dementia are lacking.
Policy implications
While the IAT ecosystem is expanding rapidly and creating novel opportunities for technology-assisted dementia care, policy makers must seek to harmonize such developments and remove or prevent administrative, regulatory, and infrastructural obstacles that could delay the integration of IAT into standard care. In addition, they also have the responsibility to address situations and conditions that could potentially undermine the successful and ethically appropriate adoption of IATs among end-users. As recently addressed by the Working Party on Biotechnology of the Organization for Economic Co-operation and Development (OECD), the global challenge of AD and other dementias requires the development of a multi-national plan that could harmonize technology development, facilitate the process of technology transfer, establish a framework for public-private partnerships for innovative projects, and create new models for multinational governance [52].
Since the increasing availability of IATs is disproportionally exceeding the number of tools currently used in clinical practice, a more effective commitment to accelerating translational research and pioneering responsible adoption is highly needed. Technology transfer is paramount to address this challenge. There is an urgent need for accelerating the translation of clinically effective technological innovation into clinical and commercial applications. This transfer will require cooperative work at the intersection between technology development and healthcare, the creation of multidisciplinary platforms for information exchange, and increased investments for innovative research as well as product development and marketing. To favor such cooperation, increased interaction between manufacturers and clinicians is required; with the former taking into account more closely the clinical needs of their end-user populations and the latter increasing their awareness of available technological applications. With the number of IAT prototypes more than doubling every five years, clinicians should monitor this rapidly expanding realm of assistive solutions, keep track of the novel technological availabilities, and supervise their responsible implementation. At the same time, manufacturers should be incentivized to adapt new prototypes more closely to the needs of patients, involve them constructively and in a participative manner into the design of future products via UC design and seek clinical validation of their prototypes through clinical trials run on larger sample sizes. Clinical effectiveness is a critical factor not only for technology adoption but, most importantly, also for guaranteeing the efficacy of technological products to improve care. Feasibility and usability tests via simulation are critical preliminary indicators of successful implementation. However, in absence of large-scale, well-designed and statistically significant clinical trials, the clinical effectiveness of IATs cannot be presumed nor generalized across various geographical or clinical contexts. In particular, studies that evaluate IAT-derived functional improvements in patients via randomized controlled design are essential. In addition, there is still a great need for outcome studies on the effectiveness of IAT interventions in non-institutional real-life settings, such as studies with home-dwelling older adults with dementia during the completion of ADL tasks. Responsible translation in IAT for dementia should follow an evidence-based strategy that prioritizes systems with demonstrated clinical effectiveness and facilitates their responsible introduction into care.
In parallel, health policies and business strategies that promote and facilitate the responsible uptake of IATs must be encouraged to prevent this reportedly large technological potential from remaining underused. To this respect, the significant increase over time of UC approaches shows that a first step in this direction is already being taken. Since the delayed transition to UC approaches to technology design and assessment has often been recognized as one of the major causes of the lower-than-expected uptake of IATs for dementia [53, 54], the increasing prevalence over time of such approaches reveals an ongoing transition in product development that is likely to ultimately result in increased societal adoption. UC approaches, in addition, are predicted not only to increase technology acceptance, but also to reduce marginal risks, increase effectiveness and maximize benefits for the end-user population [14, 23]. As UC approaches give extensive attention to the needs, wishes, and limitations of end users at each stage of the design or assessment process, IATs will be increasingly more capable to match the needs, wishes, and limitations of people with dementia. This can also result in promoting the autonomy of end-users since they are given an active, prototype-shaping role, becoming co-developers instead of being simply considered passive users of predetermined artifacts.
While AI-modulated human-to-machine interaction is a critical pathway for empowering older adults with dementia and overcoming social isolation, human-to-human interaction should be pursued too. The need for more AI-modulated human-to-human interaction is particularly important for older patients who are not digital natives.
At the ethical and social level, a number of considerations must be included into the design of new products to guarantee responsible and successful development. While the often stated goal of IAT researchers is to maximize older adults’ capacity for independent living and delay the need for institutional care [55-57], issues of privacy and information security should also be early considered during product development. Because various types of IATs could be used to access private and sensitive user-related information [58-60], privacy and security breaches should be anticipated and prevented. With security by design being hard to achieve, measures for securing sensitive (e.g., behavioral, personal, or physiological) information should be implemented at the level of product development, institutional use as well as in-home use.
Finally, cost-related and access-related considerations should be addressed to avoid the risk that IAT adoption will be impeded by socio-economic factors or could even exacerbate existing socio-economic problems. To prevent this risk, the massive adoption of IATs for the aging population should be coordinated with health policy plans and health insurance programs to minimize the emergence of adverse unintended societal consequences. For instance, reimbursement plans, government incentives, and the promotion of low-cost and open-sourced IATs are crucial strategies to promote access to technological innovation and avoid the emergence of a digital divide between older adults with dementia who could afford IATs and those who could not. Such a divide, in fact, could exacerbate existing socio-economicinequalities [61].
Conclusion
Intelligent technology is reshaping the world we live in. Its application into dementia care has a great potential for older persons and our society. Based on our systematic review, we produced a comprehensive and up-to-date index of IATs developed for assisting older adults living with dementia. This index provides health professionals with a comprehensive and up-to-date picture of the current availabilities of IATs for dementia, their major trends, limitations and possible applications into dementia care. As technology is rapidly advancing, future research should closely monitor this rapidly expanding technological spectrum and more extensively test their clinical effectiveness. In parallel, healthcare services and policies should keep up with advancing technology and facilitate the successful adoption of IATs into standard care in a manner that benefits patients, their caregivers, and the society.
DISCLOSURE STATEMENT
Authors’ disclosures available online (http://j-alz.com/manuscript-disclosures/16-1037r1.
Footnotes
Devices with own computational capacity are those that enable the collection, processing, and transfer of information without external support.