For the current proof of concept pilot study, we improved the English version of the SG4D “MBW” and built a user management and data collection system around it. We used the development process theory described at length previously (Tziraki et al., 2017), aiming to create a theory-based SG4D, with input from a multi-disciplinary team familiar with aging, dementia, user experience design, gaming theory, and technology, as well as direct input from end users (using the iterative process, see Valdez et al., 2015) and data collected automatically by the system.

In the second stage of development of “MBW” we focus further on user interface and design, aiming to increase social and emotional engagement—the act of being occupied or involved with external stimuli. It is not always clear how to effectively engage PwD in activities and derive the desired benefits, as well as how to measure engagement, especially since there has been a dearth of research examining engagement in PwD (Trahan et al., 2014; Perugia et al., 2018). There is evidence though, to suggest that interventions that support the sense of self, by using tasks or objects with specific meaning to the person, are more likely to engage PwD (Cohen-Mansfield et al., 2009; Zimmerman et al., 2011). Thus, explicit models of emotional identity offer an attractive mechanism for developing more appropriate and effective technologies. This is of specific importance, as PwD are much more focused on emotional and social issues than on abstract problems (Carstensen and Mikels, 2005; Mikels et al., 2005; Blanchard-Fields, 2007). But how exactly does one turn a flat screened tablet into a serious game that can evoke emotions and social interaction in PwD? This is truly challenging.

The methodology that we followed began with the choice of a multidisciplinary team familiar with aging, dementia, user experience design, gaming theory, and technology. Multidisciplinary collaboration is often encouraged in user centered design (UCD) including partnerships between designers and clinicians (Robert et al., 2014; Cornet et al., 2020). The synergies of the team helped to translate appropriate game ideas that came from the aging and dementia experts into emotionally and socially stimulating experiences designed with input of the user experience (UX), gaming and technology experts. We gave special consideration to design aspects that would trigger the PwD emotionally and socially, keeping in mind the importance of emotional alignment (Ienca et al., 2017; Robillard et al., 2018). We used the inclusive participatory design method which ensures benefits to end-users, promotes engagement and ensures usability (Robillard et al., 2018). We also considered more personalized aspects of the game screens (i.e., culturally appropriate, clear, welcoming pictures and design choices to enhance PwDs engagement with the SG4D).

Great care was devoted to the choice of each picture used, in order to increase emotional and social interaction with the SG4D. First, we chose to use real life photos as opposed to animations to make them more realistic and understandable. We chose pictures that would be familiar to a broad range of cultures, not childish, but that includes pictures of children, and all the generations. These intergenerational and family images with clear unambiguous emotional expressions were chosen to make it easier for PwD to recognize the emotions of the characters and empathize, sympathize and interact with them (McAllister et al., 2020), and give a feeling of warmth and belonging (see Figure 2). Nature scenes were included as well, to elicit emotions of calmness and enjoyment (Gamble et al., 2014; Thake et al., 2017).

The interdisciplinary team worked together to ensure instructions are clear, friendly and inviting for a person with dementia (Di Rosa et al., 2014; Mondini et al., 2014; Lam et al., 2015). Another aspect included throughout was appropriate and respectful humor (Berenbaum, 2003). Being able to understand humor is one of the capacities remaining for many in the middle stages of dementia. Since most people are not able to laugh spontaneously, the SG4D acts as a stimulus to trigger laughter. Indeed, laughing has positive psychological, physical and social benefits. It releases tensions, dilutes stress, and induces relaxation (Eng, 2000). Throughout, game tasks were designed to match the abilities of PwD in order to increase self-efficacy (Tziraki et al., 2017).

Each game screen was person-centered (Kitwood, 1997), and culturally appropriate to engage them in an interactive way. Figure 1 is a depiction of a sample game screen where the PwD was instructed to follow written and oral instructions to sort, find, drag, and move items on the tablet touch screen. Game screens targeted the following skills: eye hand coordination, language skills (reading, comprehension), understanding and following instructions, praxis, memory, sustained attention, and object recognition. They also targeted other physical, cognitive and emotional social skills such as gnosis, association, mental rigidity, recognition, abstraction, association, inhibition, basic math skills, logic, language and communication responses. All tasks included in game screens were in order to maintain appropriate challenges and spark interest throughout gameplay.

The system was designed in a modular way so that the game screens sequence can be interchanged, and various timings in the game are configurable. This allows for testing a variety of game experience scenarios.

The game's length, interaction timers and difficulty levels are all configurable in the system and may entail further experimentation. Indeed, the design, playfulness, feel, and looks of the system and game, were all inspired by the field of dementia care, as well as input of the MELABEV staff. We had a co-design approach to the game development based on the clients' feedback in terms of usability, and clarity for each screen of the game. We also included a training session for the PwD, before commencing the pilot, so they could learn how to use the technology. At the end of the iterative development stage, we had a prototype of a tablet-based game for PwD with 23 game screens including 15 “physical interaction screens” —where the PwD must sort, find, drag, and move items on the tablet touch screen, and eight “sensory interaction screens” —not requiring physical interaction with the tablet but where the PwD looks at pictures on the screen and interacts sensorially, emotionally and socially. This prototype was used for the pilot proof of concept study described here.

The design of the novel “MBW” SG4D can be examined using an adaptation of NASA's Technology readiness levels (TRL) scale, similar to the EU's adaptation of the scale methodology (Earto, 2014). We also look at the methodology described by a publication by the LEITAT organization (Jamier et al., 2018). MBW's readiness level can be described by the matrix (see Table 2) using a similar adaptation to the TRL scale, as described by the LEITAT organization. We denote five components of the system, aiming to ultimately provide a competitive marketable solution, which encompasses the many aspects of therapy and user interaction described by this article, is robust and scalable, and is also coupled by credible research data and analysis.

Our (1) basic unit is the game itself, an App which can run on a tablet and shows the various game screens in succession. Next, we denote the (2) Integrated System, coupling the game App with a user login base and a server back-end to collect data. All development and coding having to do with the accompanying research are denoted (3) Sub System. Lastly, the penultimate and ultimate components are the (4) Operational System, a market deployment ready prototype, and (5) Application, the final software deployment solution, a robust and scalable technology product, encompassing all stages of physical site deployment.

The different components of the system are at various readiness levels ranging from “technology concept formulated” (TRL 2) to “technology demonstrated in a relevant environment” (TRL6; i.e., an active dementia day care center) as seen in Table 2. The basic unit and integrated system of “MBW” (column 1, 2 of Table 2) have been demonstrated to be at TRL level 6. We have successfully completed an initial system pilot, testing the various aspects of the system in one of MELABEV's 4 day-care centers, which can be regarded as a relevant environment for this game.

The present proof of concept pilot study took place at MELABEV, an English speaking, dementia day center, Jerusalem, Israel (Berenbaum, 2010). MELABEV has 4 day-care centers attended by ~500 PwDs, ranging from people with moderate cognitive impairment (MCI) to advanced dementia. While the literature reports on it being difficult to engage PwD in activities, the activity staff at MELABEV do their utmost to find the appropriate activities for each client (Berenbaum et al., 2017a,b). MELABEV's professional staff routinely uses computer games on a one-to-one basis for cognitive stimulation gaming (Berenbaum et al., 2011), as well as reminiscence therapy at the computer (Douglas et al., 2004). Primary family caregivers who enroll the PwD in the day care program consent to the participation of their family member with these kinds of technology, as well as all other activities in the day care center.

Meaningful informed consent for people with dementia is challenging. Thus, for our pilot study, we utilized the participatory consent process (Dewing, 2007, 2008). Before each gaming session, the research assistant asked the PwD if he/she agreed to participate in the gaming session. Upon agreement, the PwD voluntarily got up and was guided by the research assistant to a designated space (one of four quiet rooms) to play the SG4D. If the PwD did not agree to participate, or if he/she said or acted as if he/she didn't want to continue, the game session was promptly terminated, and the PwD was led to the regular activity room at the center, with no consequences whatsoever to the services they received in the center.

Twenty-seven PwD, clients at the MELABEV dementia day center, were recruited to the proof of concept study. During recruitment, we excluded clients with aggression, delusional behavior, a history of alcohol or substance abuse, depression, severe auditory, and/or visual or motor deficits, as assessed by the professional staff at MELABEV. From the 27 who started the pilot, eight were removed from analysis for the following reasons: (a) reached ceiling in performance (the task was apparently not challenging as evident in their higher scores on cognitive assessment, N = 2); (b) did not complete more than one session (N = 4); (c) extremely low affect (N = 2). Performance of nine-teen participants was analyzed (14 women and five men, ages ranging from 65 to 97 years; M age = 87 years, SD = 8 years). Fifty-six percent were from the US, 26% from Eastern European countries, 9% from South Africa, and 9% from England. The participants had worked in various professions in the past including: programing, beautician, architecture, accounting, social work, teaching, and business.

The 19 participants included in the analysis came from two different groups at the Melabev center: Higher Functioning (13 participants) and Lower Functioning (six participants). PwD are placed in these groups based on various criteria set by the Melabev staff, which include group dynamics as well as cognitive abilities. To assess participants' preserved cognitive abilities, Melabev center uses the Montreal Cognitive Assessment Scale (MoCA; Wallace et al., 2019). Participants in the Higher Functioning group scored between nine and 28 on the MoCA scale, and participants in the Lower Functioning groups scored below six.

Ten researchers (all graduate or undergraduate interns from the fields of psychology, nursing, social work, pharmacology, pre-med, and speech therapy), who were not involved in the development process of the game, participated in the proof of concept study. They accompanied the PwD during game play. Their main task was to observe the sessions and document the PwDs engagement with the game, focusing on emotional, and social abilities. During each 20–30 min gaming session, the research assistant, working one-on-one with each client, filled in two different assessment forms (as described in the Tools section). Researchers were also instructed to help with any technological issues that might arise during game play (i.e., no internet connection; Manera et al., 2017).

Attention was paid to tablet set up before each session. The sound level was set to the client's most comfortable level, noting the impact of auditory sensory degradation in aging (for example, see Ben-David et al., 2016). Before starting, each PwD was reminded by the researchers to use only one hand on the tablet and to keep the other on the table or on their lap. Research assistants were encouraged to let the client interact with the game independently and not complete the tasks for them.

Training. Before commencing with the pilot, each PwD was presented with a tutorial on how to use the tablet properly (Cunnah et al., 2019). The research assistant informed the PwD:

Following instructions, participants commenced with the game session. In each gaming session, the PwD could play with all 23 game screens. After the completion of each game session, the PwD was asked for their feedback which they gave both verbally to the researchers and by entering it into the tablet. When mistakes (or inactivity) were recorded by the system, it provided cues/prompts to guide the participants toward expected activity (usually, by repeating the instructions). After two cues, participants were shown the correct answer, and the game advanced to the next game screen. The duration of each game session was between 20 and 30 min. All sessions took place at approximately the same time of day in a quiet room. In every game session, each PwD had the opportunity to play the complete game of 23 game screens.

All researchers described their observations, during the game play, on an ESI evaluation form, designed especially for this study. This ESI form, presented in Appendix A, was modified from the ACIS (Assessment of Communication and Interaction Skills, used by occupational therapists) (Forsyth et al., 1999; D'Amico, 2017) by a multidisciplinary team (gerontologist, MD, OT, speech therapist, psychologist). The ACIS is a structured observation tool with 20 discrete skills concerning physicality, information exchange and relations. The ESI tool is designed to capture, in detail, a person's social interactional ability while he or she is participating in a meaningful social context. It identifies strengths as well as problem areas. We included social interactional abilities from the ACIS that were feasible for the research assistants to observe during game play and that corresponded to three components of the Bar-On emotional social intelligence model: interpersonal, intrapersonal, and adaptability (Birks and Watt, 2007).

The “MBW ESI Evaluation Form” includes 11 parameters related to communication and interpersonal social skills rated on a 5-point Likert scale. If the research assistants witnessed any of the following 11 capacities in the reactions of the PwD to the tablet during any of the game screens they would rate it from 1 to 5 (one being the lowest frequency, five being the highest). The form also included one other parameter. This was filled in if at any time during the gaming session, the PwD initiated communication with the research assistant who sat next to them throughout the gaming session. This was rated on a 5-point Likert scale. This form was filled out by the researchers as the PwD was using the game. A total ESI score was derived, in order to assess reliability of the ESI Form, we conducted a Cronbach's Alpha test which indicated that the evaluation has an acceptable reliability (α = 0.92).

In addition to the ESI form, the “MBW Game Screen Subjective Evaluation Form” (see Appendix B) was used which prompted researchers to report on their subjective feelings related to the PwDs' performance on each of the 23 game screens during each game play session. The form included both close and open-ended questions, such as: (1) “Did the client enjoy the game?” (2) “Document specific quotes that the PwD shared during the game session”, respectively. The open-ended part of the form had no specific guiding questions.

A login-based system was used to monitor game plays by the PwD and saved on a centralized anonymized database for analyses. The analysis in the current study focused on “physical interaction screens.” These are 15 of the MBW game screens, where the PwD was asked to actively engage with the game—e.g., choose a specific object on the screen and drag it to a designated place on the screen.

Accuracy of performance was found to be very low in the first three sessions, as expected. Likewise, the time users spent with each game screen varied significantly between participants, which can be explained by the inability of the PwD to focus on a specific task. As a result, an aggregate “correct engagement” score was calculated. Specifically, in each game screen, choosing and/or dragging the correct object was considered correct engagement with the tablet, irrespective of success in the game screen. The maximum score for each session was 15 (as 15 physical interaction screens were analyzed) and minimal score was 0.

Subjective observations and quotation data from MBW game screen subjective evaluation form, was collected on an Excel document and analyzed using grounded theory (Chun-Tie et al., 2019). Five different researchers analyzed the data for themes, three are co-authors on this paper, and the other two had not been involved in the game development or pilot study. The group spanned different fields of expertise (medicine, occupational therapy, gerontology, social work, and neurobiology) and levels of expertise (from interns to emerita). Each researcher read the Excel data a number of times on their own searching for themes. Then they discussed all the themes found by all the researchers with about 92% of original agreement between them and came to a consensus on the themes. Then two of the researchers returned to the data to collect the quotes they felt best portrayed the themes that had been chosen that related to emotional and social intelligence. They came up with a long list of quotes which together they narrowed down to about 20 quotes for each sub-category of the Bar-On Model of Emotional Social Intelligence (total of 180 quotes). This list of quotes was given to four different professionals (three social workers and a speech-language therapist) who were asked to, each on their own, choose the three most representative quotes from the list of 20 demonstrating each sub-category of the Bar-On ESI. The quotes that best represent the different themes are presented in Table 3.

Symptoms of dementia appear to cause difficulties for the PwD when playing the SG4D. Some issues that arose were related to disorientation, distractions in their surroundings, stress, and slow speed of processing related to figuring out answers and implementing them by dragging on the screen. Motor impairments, such as physical rigidity, poor fine motor skills and sleepiness, were also mentioned as a factor. The types of mistakes mentioned by research assistants were related to mental rigidity (i.e., mentally stuck on the previous screen), apraxia (inability to initiate drag movement), poor spatial orientation (i.e., dragging in wrong direction), disinhibition (i.e., circling aimlessly on the screen), and impaired abstraction (i.e., understanding instructions literally).

The research assistants were also clear in their reports on demonstrated learning by PwD. They also reported that in 89% of the game plays participants appeared to enjoy playing the game. Some were even able to anticipate the instructions to the game, after practice. Observations indicated the use of various learning techniques in order to complete the game. Some of the PwD used the trial and error method. It was sometimes evident that they used a self-correcting mechanism when they felt that their first attempt was incorrect. Others asked the research assistants for help with specific tasks, or with their general confusion. Some reread the instructions out loud or to themselves and consciously contemplated the instructions. For example, sometimes they counted out loud to keep track of what they were doing.

In our grounded theory analysis of the research assistants' written observations of the PwD playing the SG4D, and their quotes, we found the following 9 sub-components of the Bar-On Model of ESI, as divided to three categories: (1) Intrapersonal: self-regard, emotional self-awareness, assertiveness, self-actualization; (2) Interpersonal: Empathy, social responsibility, interpersonal relationship; and (3) Adaptability: Reality testing, Problem solving. Table 3 shows selected quotes as examples demonstrating each sub-component: (words in italics are to explain the context to the reader).

In sum, the qualitative analysis suggests that PwD encountered difficulties related to dementia in playing the game. However, they demonstrated the ability to overcome these difficulties, to learn new techniques and to interact in a meaningful emotional and social way with the research assistants.