The Efficacy of Virtual Reality-Based Cognitive Training in Older Adults Non-Demented and with Mild Dementia: an One Group Pretest-Posttest Design

Background Current methods of cognitive interventions for older adults are increasingly employing modern technologies. However, research into possible applications of virtual reality (VR) in such interventions has begun only recently. The aim of the study was to evaluate the efficacy of VR-based cognitive training for older adults and to compare this efficacy in older adults without cognitive impairment and with mild dementia. Methods The complete data were obtained from 99 individuals aged 60-89, qualified according to Mini Mental State Examination (MMSE) scores into the group of non-demented older adults (n=72, mean age 68, MMSE: 28-30) or older adults with mild dementia (n=27, mean age 72, MMSE: 19-23). An one group pretest-posttest design was applied. Cognitive training with the use of GRADYS software – a computer game with elements of VR was introduced in all participants. The game included four modules corresponding to: attention, memory, visual processing and language. The intervention lasted 4 weeks and consisted of eight sessions, two per week. The intervention program was preceded and followed by a cognitive assessment of processes corresponding to the modules of the game. The following statistical tests were used: the repeated measures multivariate analysis of variance supplemented with one-way tests, the Hotelling T2 test, the Student’s t test, the Friedman test, the Mann-Whitney U test, the Kruskal-Wallis test. Results Both research groups demonstrated progress in the training, which was greater in non-demented older adults for the majority of cognitive modules except for the attention module. There were also significant differences in general cognitive functioning before and after the training. However, positive changes in cognitive performance in particular tests were revealed almost exclusively in the group of older adults without cognitive impairment. There was no relationship between the level achieved in particular training modules and the


Abstract
Background Current methods of cognitive interventions for older adults are increasingly employing modern technologies. However, research into possible applications of virtual reality (VR) in such interventions has begun only recently. The aim of the study was to evaluate the efficacy of VR-based cognitive training for older adults and to compare this efficacy in older adults without cognitive impairment and with mild dementia. Methods The complete data were obtained from 99 individuals aged 60-89, qualified according to Mini Mental State Examination (MMSE) scores into the group of non-demented older adults (n=72, mean age 68, MMSE: [28][29][30] or older adults with mild dementia (n=27, mean age 72, MMSE: 19-23).
An one group pretest-posttest design was applied. Cognitive training with the use of GRADYS software -a computer game with elements of VR was introduced in all participants. The game included four modules corresponding to: attention, memory, visual processing and language. The intervention lasted 4 weeks and consisted of eight sessions, two per week. The intervention program was preceded and followed by a cognitive assessment of processes corresponding to the modules of the game. The following statistical tests were used: the repeated measures multivariate analysis of variance supplemented with one-way tests, the Hotelling T2 test, the Student's t test, the Friedman test, the Mann-Whitney U test, the Kruskal-Wallis test. Results Both research groups demonstrated progress in the training, which was greater in non-demented older adults for the majority of cognitive modules except for the attention module. There were also significant differences in general cognitive functioning before and after the training. However, positive changes in cognitive performance in particular tests were revealed almost exclusively in the group of older adults without cognitive impairment. There was no relationship between the level achieved in particular training modules and the improvement in corresponding cognitive tasks. Conclusions Currently, we can recommend the GRADS game for use only in non-demented older adults. The hypothesis of the presumable usefulness of the game in individuals with MCI needs to be tested. In turn, the application of the GRADYS game to people with dementia would require modification of the hardware and software. Background 4 The dynamic development of modern technologies raises questions as to the possibilities, benefits and limitations of their application among the older adults due to the developmental changes in the sensorimotor abilities, cognition and motivation observed in this age group (1). Although seniors proved to be late adapters to the world of modern technologies, the fact remains they effectively use those technologies for their own health protection and promotion, for instance telemedicine, e-health (2).
An important area for modern technologies application in older adults is improving their cognitive functioning through cognitive interventions (CI), including cognitive training (CT), cognitive stimulation (CS) and cognitive rehabilitation (CR) (3). As people get older, many cognitive abilities deteriorate and the process is referred to as cognitive ageing (4). It mostly affects memory processes, learning, attention, reasoning and executive functions (5,6,7). What is more, the risk of dementia disorders increases with age in the course of various neurodegenerative diseases (8,9). Deterioration in cognitive functioning of an older adults results in their reduced independence, lower self-esteem and withdrawal from many areas of activity (professional, social and educational) due to deterioration in mental functions. In view of the above, studies into possible forms of non-pharmaceutical interventions aiming to maintain, improve or enhance cognitive performance, and prevent its deterioration in senior age gain particular relevance.
Research on the methods improving cognitive performance in cognitively healthy older adults and in persons with cognitive impairment, including dementia, have been conducted for many years now and have provided evidence of the effectiveness of various cognitive interventions (10).
For example, the meta-analysis of randomized controlled trials by Chiu and colleagues (11) indicated that CT is effective for non-demented older adults. Similarly, the review of research on CS and CT programs for healthy older participants performed by Tardif and Simard (12) pointed out that the results are promising for memory, attention, executive functions, and speed of processing. Also metaanalyses of the influence of process-based CT, like working memory or cognitive control training, in healthy older adults demonstrated the effectiveness of such CI (13,14). Moreover, CT enhance stability of cognitive functioning across adulthood, as indicated the review by Eschen (15). According to this review all types of CT enhanced absolute stability of cognitive functioning, but the greatest effects were reported for process-based CT. What's more, findings of a few studies discussed in this review demonstrated that thanks to CT also the absolute stability in brain functioning across adulthood can increase. Also systematic review of randomised controlled trials with longitudinal follow-up revealed that CT can prevent the onset of dementia in healthy older adults (16). CT induced strong and persistent protective effects on longitudinal neuropsychological performance. Some studies considered in this review reported also transfer of training effects to general cognition and daily functioning.
Moreover, CI may be beneficial not only in normal ageing process, but also in the case of MCI. The results of all CI in cognitively healthy older adults and older adults with MCI between 1970 and 2007 reviewed by Martin et al. (17) suggest that CI lead to performance improvements and that the size of the effects differs for different kinds of memory skills in non-demented older adults and people with mild cognitive impairment (MCI). Systematic review of 10 studies, followed by effect sizes analysis found that CT (including cognitive exercises and memory strategies training) can produce moderateto-large positive effects on memory-related outcomes in older adults with MCI (18). Positive influence of CI, including CT, on various aspects of memory, attention and executive functions in older adults with MCI was pointed out also in other review (19). Mewborn, Lindbergh and Stephen Miller (20) analysed 279 effects from 97 studies on CT in older adults. Overall, results indicated that CT produce a small, but significant, improvement in the cognitive functioning of older adults, relative to active and passive control groups. At the same time, cognitive status (cognitively healthy vs. MCI), as well as age and education were not significant moderators. Effects were larger for directly trained outcomes but were also significant for non-trained outcomes (i.e. transfer effect). However, not all reviews and meta-analyzes on the effects of CI in individuals with MCI lead to such optimistic conclusions. A meta-analysis on memory training effects pointed out the scarce of evidence of the effectiveness and specificity of such CT in older adults without cognitive impairment and with MCI (21).
Going further, the results of research on the possibility to improve cognitive functioning under the influence of CI in persons with dementia are less clear. Some reviews and meta-analyzes indicated positive effect of CI on cognitive functioning in dementia (22,23,24). Based on the analysis of randomized controlled trials, Ballard and colleagues concluded, that modest but significant benefits in the treatment of cognitive symptoms in people with Alzheimer's disease (AD) can be achieved through various CI, including CT, CR and CS (25). Also the systematic review of the literature and meta-analysis of the effect of CT on multiple functional domains in AD patients showed medium effect sizes for learning, memory, executive functioning, activities of daily living, general cognitive problems, depression, and self-rated general functioning (26). According to the scoping review on the effects of non-pharmacological interventions for adults with mild cognitive impairment and early stage dementia, CI focused on remediation caused a little improvement in selected cognitive abilities, in turn cognitive training focused on compensation influenced the impact of cognitive changes impact on daily living (27). The effect sizes were however small. Moreover, most of the studies included in this review concerned MCI patients. The review comprised 20 studies in persons with MCI, only 8 studies in persons with early AD and 4 studies involved both, MCI and early AD patients. Other authors pointed out that convincing evidence of clinical significance for the impact of CI in cognitively impaired older adults was only obtained from single trials in terms of delay of cognitive decline, improvement in activities of daily living, or enhanced attainment of personally relevant goals (28).
Furthermore, if CI types are distinguished, i.e. CT, CS and CR, it turns out that in people with dementia they vary in terms of effectiveness. Meta-analysis and meta-regression by Huntley and colleagues found the significant positive effect of CS on general cognitive functioning in dementia, but no evidence for the positive influence of CT or mixed approach, combining CT and CS (29). Also according to review by Bahar-Fuchs, Clare and Woods (30), CT is not associated with positive or negative effects on cognitive functioning in persons with mild to moderate dementia. In the review on CI in cognitively impaiered older adults authors firstly concluded that CI improve global cognitive functioning in patients with MCI or AD and their abilities of daily living, reduce behavioural disturbances, and have positive effects on quality of life. Secondly, authors noticed the varying effectiveness of particular types of CI, depending on the level of cognitive deficits. Patients with mild to moderate dementia benefited more from CS, while older adults with MCI from CT (31). It corresponds to the conclusions from the systematic review of 11 CT studies and 7 CS studies in individuals with dementia (32). In this review evidence for efficiency of CT among demented persons proved to be insufficient and doubtful. Contrary, for CS there was good evidence for general cognitive enhancement, more specifically in language and memory.
Currently, an increasing number of research concerns the efficacy of CI using modern technologies, like computerized cognitive training (CCT) or video games (VG). A systematic review on efficacy CCT for cognitively healthy older adult indicated that findings are comparable or better than those from reviews of more traditional, "paper-and-pencil" CT approaches (33). The authors concluded that computerized training may be an effective alternative. Lampit, et al. (34) reviewed the results of fiftytwo studies on the effectiveness of CCT encompassing 4,885 healthy older adults. The overall effect size for CCT versus control was small but statistically significant. Small to moderate effect sizes were found for nonverbal memory, verbal memory, working memory, processing speed and visuospatial skills. No significant effects were found for executive functions and attention. According to the systematic literature review and meta-analysis by Tetlow and Edwards (35) commercially available CCT can improve cognitive abilities in older adults, who also report improvement on tasks relevant to their everyday lives. Other systematic review on clinical significance of commercially available computerized "brain training" programs was summarized by the authors' conclusion that at least some of such CT software is effective and can supporting healthy ageing (36). Moreover, Edwards et al. (37) conducted a systematic review and meta-analyses of Useful Field of View (UFOV) CT, and their results indicated that UFOV CT enhanced neural outcomes, speed of processing, and attention and showed far transfer to everyday functioning. UFOV CT effects were equivalent when compared to active-or no-contact control conditions. Moreover, improvements on the trained skills endured across ten years. Admittedly, they did not transfer to other neuropsychological outcomes, however positively enhanced well-being, health, and quality of life. A systematic review and meta-analysis on computerbased CI for people with dementia proved that these CI have moderate effects in cognition and mood but not on activities of daily living (38). A PRISMA-compliant network meta-analysis led to conclusion that in older adults with MCI and AD optimal intervention for cognitive performance is physical exercise, whereas CCT is the optimum for neuropsychiatric symptoms (39). This meta-analysis also showed that nonpharmacological therapies are better than pharmacological therapies in MCI and AD patients. Another meta-analysis concerned the efficacy of CCT in people with MCI or dementia indicated moderate overall effect on cognition in MCI and small to moderate effects for global cognition, attention, working memory, learning, and memory, with the exception of nonverbal memory (40). In turn in dementia, statistically significant effects were found on overall cognition and visuospatial skills, but the last ones were driven by trials of VR or VG. The authors concluded that CCT is efficacious in people with MCI but evidence for efficacy in people with dementia is weak and limited to trials of immersive technologies. The newest scoping review on non-immersive brain gaming for cognitively impaired older adults showed that most brain gaming interventions with the use of computer and/or touch screen led to improvement in at least one cognitive outcome (41). Going further, according to this review in older adults with MCI it was possible to get the near transfer effect by training in processing speed, memory, attention, and working memory, the far transfer effect by training in attention and working memory, and improvement in functional activities by training in visual processing speed and attention, or working memory. In turn, in older adults with dementia the supervised brain gaming training in memory, executive function, and language led to the near transfer effect. The conclusion of the authors points to the potential of brain gaming in the improvement of cognitive functioning in older adults with mild to severe cognitive decline. As regards VG, from a meta-analytic study on the enhancement of cognitive functioning in healthy older adults under influence of VG-based CT followed that such training induces positive changes in reaction time, attention, memory, and global cognition (42). The research by Wang, Zhu, Qi, Huang, and Li (43) proved that VG experiences may have a positive influence also on the brain activation underlying cognitive performance. Their results showed that healthy older VG players presented significantly better cognitive performance than non-VG players and greater brain activity, mainly in frontal-parietal areas.
Along with the intensive development of modern technologies, more and more advanced solutions are becoming available. In this context the use of VR in the CI is a new research area arousing growing interest. VR can be defined as "an advanced form of human-computer interface that allows the user to interact with and become immersed in a computer-generated environment in a naturalistic fashion" (44, p. 298). Currently the standard for fully immersive VR are head-mounted displays (HMDs) because they provide the greatest level of immersion. Besides HMD, VR can be implemented by world-fixed displays taking many forms, from a standard monitor to displays completely surrounding the user e.g., CAVEs, and hand-held displays, like smartphones or tablets (45,46).
Evidence is accumulating showing that VR-applications can successfully be employed for early detection and monitoring of physical and cognitive impairment (47,48,49,50), but also for interventions in a wide range of medical conditions, like chronic pain, obesity, eating disorders, anxiety, phobia, depression, schizophrenia, autism, behavioural disorders and cognitive disorders of various etiologies (51,52,53,54,55,56,57).
One major benefit of VR-based CT is that it addresses previous criticisms about CCT by providing individuals with an intervention that is immersive, naturalistic, and mimics real-time and real-life, increasing ecologic validity (54,58,59,60). Ecological validity provided by VR can be seen as a key component for assessing and training cognitive skills that are relevant for functional tasks in realworld contexts (61,62,63).
Moreover, it is worth to be noted that VR applications like VG and CCT create favourable conditions to implement the rules of so called serious games in CT for older adults (64,65). Serious games have been used in the field of neurodegenerative disorders such as Alzheimer's disease to support and improve the assessment of different functional and cognitive abilities, and to provide alternative solutions for patients' rehabilitation. Results confirmed that SG are also adapted to older people with MCI (66).
In summary, currently VR technology is considered as one of the most promising tools for providing nonpharmacological cognitive interventions for cognitively healthy older adults and persons with different level of cognitive impairment (54,67).
Nevertheless, research on the effectiveness of VR-based CT is still scarce, all the more so in the group of older adults and in relation to cognitive ageing. For example in the systematic review of VR applications in inpatient medical settings between 2005 and 2015, only one of 11 studies included in the review concerned cognitive rehabilitation. This study was however not focused on cognitive ageing. It involved 18 patients with traumatic brain injury, aged 19-73 years (51).
The efficacy of CCT and VR-based CT in older adults with MCI or dementia was assessed in systematic review by Coyle, Traynor and Solowij (68). The cognitive domains of attention, executive function, and memory (visual and verbal) showed the most consistent improvements, including long-term effects.
CCT and VR-based CT were however moderately effective in long-term improvement of cognition for those at high risk of cognitive decline and did not improve the activity of daily living. It should be noted that only three studies using VR-based CT were taken into account in the systematic review, although the authors pointed out that the review was undertaken following a wide search of the available literature on the topic area. It indicates the small number of studies carried out so far on this topic.
Recently, several studies have been published, the results of which are generally promising. However, the study samples are usually small, including case studies and articles describe pilot studies or even the research plan with only partial results from initial phases of the study which are continued (61,69,70,71). It is recommended that research into possible applications of VR in CI shall be conducted because of the preliminary nature of most data currently available in this area and a relatively small number of studies carried out so far (54,68).

The aims and study design
The aim of the study was to evaluate the efficacy of the VR-based cognitive training in persons aged 60 and above without cognitive impairment and with mild dementia.
In our research we have posed the following hypotheses: 1.
In the process of training using GRADYS game an improvement in the performance of the training tasks will occur, manifested in reaching higher levels of the game difficulty in each of the training modules, regarding: memory, attention, language, visual processing.

2.
The effect of the training will result in an improvement in cognitive performance beyond the game environment, i.e. in the performance of standard cognitive measures of memory, attention, language and visuospatial processing.

3.
Belonging to the group of older adults without cognitive impairment vs. older adults with mild dementia will differentiate the participants in terms of cognitive performance in pretest and in terms of training effects. The cognitive performance in pretest will be better in non-demented older adults compared to participants with mild dementia. Training effects in both groups will be significant and positive, as described in hypotheses 1 and 2, but in the group of non-demented older adults they will be greater.

4.
There will be a positive relationship between the progression in the training and the improvement of cognitive performance beyond the game environment, i.e.
participants with greater achievements in training will demonstrate a greater improvement in cognitive performance beyond the game environment.
An uncontrolled pretest-posttest study design was applied. The GRADYS game was tested through training organized for the study participants. The training was preceded and followed by the cognitive functioning measurement. For each cognitive function, the study participants started the first session at the first, lowest difficulty level. The highest possible difficulty level to achieve was level 3. The participant moved to a higher level having reached 75% accuracy in the previous training session. In the case of accuracy falling below 50%, the participant returned to the lower level of the game.
Before starting the training programme and after its completion, the study participants' cognitive functioning was evaluated.

Participants
The study was conducted in Poland. All participants were Polish citizens for whom Polish was a native  Table 2. A significant loss from the group of subjects with mild dementia resulted from difficulties faced during training, their need to be accompanied by caregivers who often showed lack of time and tiredness of study procedure leading to its discontinuation.
The study was conducted in the Institute of Psychology at the Kazimierz Wielki University in Bydgoszcz, and in the Departments of Geriatrics and Psychiatry at the Nicolaus Copernicus University in Torun, Collegium Medicum in Bydgoszcz as well as in a healthcare facility of the "Nowa Wspolna Droga" Association in Gniezno.

Intervention
The work on the development of the GRADYS game can be described based on the scheme of technology development and evaluation process (1). This scheme describes a sequence of activities that starts with the research on the needs assessment of the intended users of a given technology, and continues in two phases, the lab phase and the field phase.
The starting point for the work on the GRADYS game was to identify the needs of the end user and the nature of CT to be implemented in the game. The above-mentioned cognitive decline observed in the ageing process, supported by numerous research findings, provides objective evidence that CT is a significant and positive intervention targeted at older adults. Based on the current literature on cognitive ageing the game was decided to engage four main cognitive functions: two that show pronounced deterioration with age even in the absence of dementia (and in the case of dementia even more pronounced), i.e. memory and attention, and two that deteriorate less seriously in normal ageing but significantly in the course of dementia, namely language and visual-spatial function.
The perceived need in older adults to improve their cognitive function was confirmed in meetings with people from this age group, at Universities of the Third Age among others and at various institutions for seniors. Given the characteristics of the target group, the tasks for the game then developed were to simulate cognitive functions while maintaining the users' physical comfort and mental welfare as well as their interest and motivation to use the game. In order to realize those tasks, it was assumed that GRADYS shall meet the following criteria.
The exercises that engage cognitive functions shall be embedded in the context of real-life places, situations and every-day activities of seniors living in Poland for the scenario to be perceived as natural and familiar to the user.
The game environment shall reflect the natural environment, yet it shall not be too rich as not to provide undue stimulation that is beyond the attention and perception capacity of the end user.
The game shall enable the user to play at increasingly higher and more difficult levels and the players shall be informed about their results and progress made in the game to induce and maintain their motivation to play the game.
The game shall be played while seated to ensure the user's physical comfort and safety since VR setting prevents visual control of the surroundings.
It is necessary for a simultaneous use of game devices, including VR and standard visual and auditory correction devices that are often indispensable for older adults.
The VR devices shall be relatively easy to use so that they do not pose an additional challenge for or build resistance and fear in the end users.
The users shall be able to play the game on their own without the assistance of an instructor or any trained person for the end product to be used by older adults in their homes.
The game as a whole (both software and hardware) shall use such solutions that will make the end product affordable for institutions working with seniors in Poland, and possibly -also for individual users, being private persons.
The subsequent phases of game development in the context of the concerned scheme of technology development and evaluation process involved laboratory work and have been described in Table 2.
The final result of laboratory work was the GRADYS game which is a CT with VR elements for cognitive enhancement in older adults. The efficacy of this game as a CT was tested in the research.
There are four modules in the game dedicated to the four areas of cognitive functioning: (1) attention; (2) memory; (3)

Outcome measures
Two sets of research tools were used: screening tests that help select subjects for the study sample and tools that evaluate cognitive functioning before starting the training with the GRADYS game (pretest) and following its completion (posttest).
The following research tools were used in the study:

Data analysis
The comparison of baseline cognitive performance in non-demented older adults and older adults with mild dementia was made using the Hotelling T 2 test -for cognitive performance in general and the

Results
Baseline cognitive performance. As expected, statistical analyses indicated better baseline cognitive performance in the group of non-demented older adults in general (Hotelling T 2 = 130.868; p < .001) and in the majority of pretest cognitive measures (Table 3). In a few of cognitive measures, there was no significant differences between groups ( Table 3). All these cognitive measures was relatively easy. Surprisingly, two performance indices in the sustained attention task (d2) were significantly better in the group of older adults with dementia in comparison to participants without cognitive impairment (Table 3). However, the WZ-B indicator is vulnerable to overestimation due to the skipping a part of the characters in the row. Therefore, the high WZ indicator can be an artefact resulting from the fact that participants with mild dementia often acted contrary to instructions of d2 test to follow the letters in a row one by one and moved immediately to further letters. The value of the next indicator higher in the group with dementia, i.e. WZ-B, can be a derivative of WZ value.
The progress throughout the training. Both groups demonstrated progress throughout the training: older adults without cognitive impairment and older adults with mild dementia ( Table 4, the Friedman's test; Figure 2). Yet the group of non-demented older adults showed greater progress for the majority of cognitive modules except the attention module, in which the difference between the groups in the last training session was not statistically significant ( Table 4, the Mann-Whitney U test).
At the same time, the mean and the median values of the difficulty levels achieved by the participants in consecutive training sessions, as well as the percentage of participants who achieved particular levels of difficulty in subsequent sessions of a given module indicated uneven difficulty of the cognitive modules. The memory module was found to be most difficult for the study participants, while the attention module was the easiest one.

Changes in cognitive functioning under the influence of training with the GRADYS game.
RM MANOVA with group as a between-subject variable showed two significant effects (Table 5).
Firstly, the groups of older adults without cognitive impairment and older adults with mild dementia differed significantly as regards the level of cognitive functioning. Secondly, there was also significant differences in cognitive functioning before the training with the use of the GRADYS game and after its completion.
As regards the interaction effect, it is not significant yet the probability value (p = .073) was only slightly above the assumed value of statistical significance with α = .05. While partial eta squared for the interaction effect was 0.348 which means a large effect size (73). This finding is sufficient to justify further analysis that help to isolate exactly where are the significant means differences.
Obtained effect size suggested that the group may be a factor moderating the effect of training with regard at least to a part of applied measures of cognitive abilities.
According to RM MANOVA computed for each group separately, significant changes occurred only in older adults without cognitive impairment (Table 6). However, in the group of older adults with mild dementia the observed power was low, in turn the effect size was big. It suggest that the statistical power of conducted test was insufficient to estimate the cognitive changes after the training in the population of older adults with mild dementia. The reason probably was, at least partially, too small sample size and high sampling error.
In turn, one-way tests for the differences between pretest and posttest for particular cognitive measures computed in both groups indicated some significant changes in the group of older adults with mild dementia, however much less than in the group of non-demented participants (Table 7).
Cognitive domains for which the improvement proved to be most pronounced were attention and visuospatial processing.  (Tables 8 and 9). These comparisons were conducted separately in non-demented older adults and older adults with mild dementia. In some cases, three groups were compared, corresponding to three levels of difficulty that could be achieved in the last training session (Table 8).
Ultimately, in some cases only two groups were compared, as in cases where very few participants remained at the lowest level of difficulty at the end of the training (level 1), this subgroup was combined with participants that have reached level 2 (Table 9). This was only the case for the older adults without cognitive impairment. No comparison of the training achievements in the attention module was made because almost all participants finished this module at the maximum difficulty level. It was 67 participants form a 72-strong group of non-demented older adults (96%) and 23 participants from a group of 27 older adults with mild dementia (85%).

Discussion
The aim of the research discussed in the present paper was to evaluate the efficacy of the cognitive training using the game with elements of VR in the target group of people aged 60 and above without cognitive impairments and with mild dementia.
The study leads to the general conclusion that the GRADYS game can be used by non-demented older adults. Whereas the usefulness of the GRADYS game for people with mild dementia is questionable and at this point we cannot recommend our software for this group.
Participants with mild dementia before the start of the training were worse in cognitive tasks than non-demented older adults, which is a rather obvious result. Nevertheless, subjects from both groups showed progress in the course of training within all cognitive modules included in the software. The progress was indicated by increasingly higher difficulty levels reached by the players in subsequent training sessions. One can therefore conclude that the proposed solution is learnable and effective when it comes to improving on the performance of the trained cognitive tasks for all participants of the study. However, among older adults with mild dementia not only less progress in the training, but also a large loss of participants from the sample was observed. Many subjects from the group of adults with mild dementia selected for the study withdrew from participation during the training process, some of them in view of major difficulties involved in its completion, including difficulties involved in controlling the game and inability to concentrate on cognitive tasks due to an excessive cognitive burden in using new technology (getting their bearings in VR, using controllers). In those cases, low learnability caused lack of acceptability of the GRADYS game, which in turn contributes to decreased efficiency of this game in older adults with mild dementia. In the case of older adults without cognitive impairments, we can speak about overall enhancement of attention indicators. In this group also improved: visuospatial processing (although only two its indicators were taken into account in the study), visual aspects of memory and working memory, but not verbal learning and language.
The situation is quite different among older adults with mild dementia. Positive Poor training effects in participants with mild dementia, despite its higher effectiveness in nondemented older adults are consistent with the results of previous studies, which prove that in individuals with dementia CS or perhaps CR work better than CT (22,25,29,30,31,32). At the same time, research shows that in the case of MCI, the situation is reversed and CT is more effective than CS (31). Moreover, in persons with MCI CT leads to better effects than in those with dementia (40) and can even retain the effectiveness comparable to those observed in older adults without cognitive impairment (20). Since the CT using GRADYS game showed sufficient learnability for non-demented older adults and some older adults with mild dementia, it can be assumed that it may also be used by individuals with MCI. This hypothesis would require an empirical verification, however, the results of previous research encourage testing the effectiveness of the GRADYS game in MCI patients.
Other conclusion regards possibilities to enhance the game through making the game control more natural and easier by, for example, replacing the control pad with a haptic glove for VR. Thanks to such improvements the game may become useful also for older adults with mild dementia. The shortcoming of our research was also design a game control system too difficult for participants with mild dementia causing them to withdraw from training.
The limitations also include the unequal number of neuropsychological measures of particular cognitive functions with fewer indicators for language and visuospatial functions.

Conclusions
The GRADYS game can be useful in cognitive training in non-demented older adults. The hypothesis that it can also be used in older adults with MCI needs to be tested. Currently, we cannot recommend the GRADYS game for use in persons with dementia, but maybe it will be possible after the adaptation of software and hardware of the game. The rapid development of modern technologies creates opportunities for improving current conceptual solutions, software and hardware with regard to the game natural feel, intuitive simplicity and comfort, as well as its affordability for institutions working with seniors and individual users. This article does not contain any studies with animals performed by any of the authors.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.   Table 2 GRADYS game development and evaluation process in the laboratory phase, based on the scheme proposed by Schulz et al. (2015) Steps/Stages (in Lab)

User-centred design prototype
Designing gameplay scenarios and cognitive tasks involved therein. Designing the game interface and solutions for the mechanism underlying interaction with the game, the navigation mechanisms, the rules of promotion to the next level in the game and how to communicate with a player and inform them about their results. Planning on a set of technical solutions necessary to navigate the game.
Storyboarding; Individual in-depth interviews; Focus group.
Robust prototype Programming the game while taking into account all the aspects mentioned above, selecting graphic environment and the assets. Testing the game software and solutions applied therein as well as technical solutions and devices used.
Laboratory testing Scenario tes Observation of task performing the team members).

Laboratory prototype
Testing the target set of software and hardware, including final game scenario as well as the mechanisms for interaction between player and game and navigation mechanisms with regard to their user-friendliness, intuitive simplicity, and learnability.
Observation of task performing the targeted end users and the team members); "Think aloud" the targeted end users); Focus group.  Note: a OA-ND -Non-demented older adults; b OA-MD -Older adults with mild dementia  Note: a OA-ND -non-demented older adults; OA-MD -older adults with mild dementia Table 7 The differences between pretest and posttest for particular cognitive measures in the groups of nondemented older adults and older adults with mild dementia. Note: a OA-ND -non-demented older adults; b OA-MD -older adults with mild dementia