Introduction

Background

Global public health in the twenty-first century faces major challenges related to the aging of the population and the expected increase in cases of dementia. According to projections cited by the World Health Organization (WHO), the world’s population aged over 60 is expected to double by 2050 [1, 2]. Aging presents many challenges, one of which is associated with age-related cognitive decline. Memory, as a cognitive process used on a daily basis, becomes an easily recognizable indicator for the detection of cognitive decline. A memory problem reflects impairments in cognitive functions such as attention, processing speed, language or executive functions [4]. Mild Cognitive Impairment (MCI) is common with age and is associated with increased mortality [7].

Cognitive Impairments may arise from various conditions, such as age-related decline in older adults. These impairments may involve deficits in working memory, which is a cognitive system that maintains, processes, and links information between perception, long-term memory, and action [8, 9]. Working memory plays an important role in executing daily living activities successfully. Moreover, executive functions, which encompass a broad set of cognitive processes affecting learning abilities, problem solving, and self-control, contribute to the execution of activities comprising daily life and are important for preserving an individual’s ability to remain independent [10].

Treatments for cognitive decline

Clinical trials have been conducted to delay cognitive decline, including cholinesterase inhibitors such as donepezil and antioxidant compounds such as vitamin E [12,13,14]. Overall, pharmacological trials in mild cognitive impairment have produced limited and inconsistent results, and some treatments may be subject to adverse effects [14, 15]. Currently, there is no specific pharmacological treatment for people with MCI, despite the condition affecting approximately 15–20% of individuals over 65 years old [11]. Non-pharmacological interventions, based on principles of brain plasticity, represent promising approaches for supporting cognitive functions. Brain plasticity reflects the ability to establish new neuronal connections and strengthen existing synaptic links in humans. This ability is maintained with age and seems to play an important role in research on mitigating cognitive decline [16, 17]. Recent research highlights the role of neuronal and cognitive plasticity for the aging brain [16, 19]. The theory of neuroplasticity appears to be an important theoretical framework for research on computerized cognitive training as it helps explain how cognitive stimulation may support adaptive changes in brain function. However, the decline with age is heterogeneous and varies among individuals. Brain plasticity is also subject to variation largely attributable to differences in cognitive reserve which, in turn, is associated with how individuals cope with cognitive decline and brain pathology. Cognitive reserve is defined as “individual differences in how people process tasks allow some to cope better than others with brain pathology” [20]. Factors influencing variation in cognitive reserve include genetics, education, occupation, socioeconomic status, physical health, lifestyle, and mental activity [21]. Individual intervention is therefore necessary in the prevention and treatment of cognitive decline since needs are variable and malliable during each intervention.

Cognitive training intervention

Cognitive training can be defined as “an intervention providing structured practice on tasks relevant to aspects of cognitive functioning, using standardized tasks, and intended to address cognitive function and/or cognitive impairment directly” [22]. Cognitive training has shown effects on working memory, processing speed and visuospatial abilities [23, 24]. However, practical difficulties limit their use, particularly with regard to face-to-face prescription, making the intervention inaccessible to patients with reduced mobility [25]. Furthermore, studies carried out on cognitive training have received criticism concerning the high proportion of results attributable to placebo effects, mainly due to the fact that it is not possible to hide the expectations of the studies. The contextual effect inherent to a training intervention and the self-selection bias of volunteers are mainly highlighted as a factor that limits the conclusions [26, 27].

Advent of computerized cognitive training

Computer technology has created new opportunities to diversify programs by moving towards computerized cognitive training [28]. Simultaneously, the advent of computerized cognitive training has resulted in industrial market growth, with a plethora of cognitive stimulation and training applications with controversial levels of evidence [29,30,31]. Nevertheless, extant research findings are in agreement regarding the modest to moderate effects of initial computerized cognitive training programs on both overall cognition, but also on more specific functions, such as executive functions, processing speed, attention and memory [11, 18, 32, 33]. Analyses suggest that computer programs may lead to improvements in symptoms of neuropsychiatric disorders, including anxiety and depression [34, 35]. Computer-based training presents new opportunities, leveraging the benefits of e-Health, such as cost savings for healthcare facilities, solutions for areas with limited medical access, and reduced intervention costs. Those technologies enable both accessible and inclusive sessions. This is achieved through the provision of at-home sessions for individuals with reduced mobility, and the incorporation of common technologies, such as tablets and smartphones [36]. Many studies have focused on engagement and motivation in computerized cognitive training for seniors during computerized programs [18, 30]. Counter-intuitively, cognitive training does not appear to require prior computer experience [37]. Digital environments provide a larger range of activities with adjustable levels of complexity, enabling the opportunity to engage in specific training targeting particular cognitive processes such as inhibition and visual attention [38,39,40]. Real-time feedback and adjustments have the capacity to enhance patient engagement and individualize remediation. Nevertheless, the effectiveness of computerized cognitive training remains a subject of controversy, particularly with regard to the magnitude of the effects and the transfer effects [41, 42]. Skepticism towards computerized cognitive remediation has been partly driven by the lack of standardization of studies, which complicates the analysis of results. In addition, variability in intervention protocols and an absence of randomized controlled trials did not provide the opportunity to draw conclusions on their effectiveness [39]. Methodological disparities in computerized cognitive program research are influenced by the potential protocol effects on the results, heterogeneity of the study population, variability of specific cognitive functions targeted, and a presumed paucity of rigorous control over socio-demographic and genetics factors [35, 43,44,45,46,47].

However, recent work has provided crucial knowledge for more standardized studies. Long-term compliance, short but frequent sessions, and favorable conditions can influence observed outcomes in computerized training [48, 49]. Gamification, defined by the use of game elements in contexts other than entertainment, is a contributing factor to adherence to programs necessitating long-term intervention [50, 51]. Playful interfaces appear to increase the attractiveness of the exercises offered in bright environments and simple interaction, maintaining patients’ motivation for the environments offered [18]. Some studies have highlighted the main factors of disengagement. The primary one comprises of the difficulty in integrating traditional activities into daily life. The second one pertains to the lack of diversity in the exercises offered [52]. To respond to these limitations and reach the public in an enhanced preventive approach, aiming to maintain cognitive faculties, a global approach to programs on smartphones and tablets appears appropriate. In addition, touchscreen technology meliorates accessibility to home programs for patients not only by maintaining communication with therapists, but also by integrating several activities into their daily routines [18]. Touchscreens are generally considered easier to use than other human-machine interface aids such as keyboards and mouse. These engaging results for tactile interfaces should be considered as they vary depending on the population studied. For seniors prone to Parkinson’s disease, in particular, studies involving fine movements on touchscreens should be avoided. Despite the computers’ predominance in computerized cognitive training studies, the trend could be reversed in the future by the increased use of tactile devices [18, 52, 54, 55]. Indeed, touchscreens appear to compensate for the limitations of other interfaces and provide a more accessible therapeutic medium for the elderly [3]. It is hypothesized that a focus on enhancing accessibility through touchscreens and gamified content can mitigate many of the challenges associated with cognitive training for older adults. Gamified exercise refers to activities that incorporate game design concepts such as user feedback or reward systems into their design.

Aims and objectives

The aim of this study is to analyze the latest advances on effectiveness of computerized cognitive training using touchscreen devices. To our knowledge, although several studies have been carried out on computerized cognitive programs [38, 56, 57], none have validated the results of RCTs on programs for the elderly specifically distributed on touchscreens. As such, no systematic review was found in the PROSPERO database. However, it is hypothesized that the attribution interface plays a major role in program outcomes in addition to facilitating standardized practices. This study was conducted with the aim to better understand the factors contributing to the divergent outcomes associated with computerized cognitive programs. It focuses on a form of program administration that assumes an improvement in results through standardization of protocols and an improvement in the accessibility of interventions for seniors.

Fig. 1
figure 1

Flowchart

Full size image

Methods

Procedure

To assess the effects of cognitive training delivered via touchscreens, a systematic review was conducted during early 2025, and early online publications available within this interval were included. This review was conducted following the main principles of the PRISMA 2020 guidelines, including a structured search strategy, predefined eligibility criteria, and systematic data extraction. It is covering randomized controlled trials (RCTs) including training exercises distributed on touchscreens and published between 2016 and 2025 in Google Scholar, PubMed, Cochrane, and ScienceDirect databases. This period was determined to encompass the latest developments in recent touchscreen interventions, as these had become widely popular among the target population during this time. The selection of articles was carried out by a reviewer (the first author), respecting the inclusion and exclusion criteria to ensure a strict and transparent process. The search string was (“cognitive training” OR “cognitive stimulation” OR “brain training” OR “memory training” OR “game training”) AND (“senior” OR “older adults” OR “elder” OR “aging”) AND (“touchscreen” OR “smartphone” OR “tablet” OR “mobile device” OR “touch”) AND (“randomized controlled trial” OR “RCT” OR “clinical trial”). Octoparse software was utilized to extract articles in tabular format and process duplicates. The language and thematic relevance of the titles were analyzed manually.

Inclusion criteria

Studies were included if: (1) Participants were 65 years or older; (2) They maintained their pharmacological treatment unchanged throughout the protocol; (3) Cognitive training was administered through touchscreen devices; (4) Published from January 2016 to January 2025; (5) written in English or French. Studies were excluded if: (1) technologies, outcomes, or methodologies were not clearly described; (2) they dealt with computerized intervention without cognitive exercise; (3) patients received pharmacological psycho-stimulants or drugs to enhance cognition.

Results

Study characteristics

Thirty-four studies, representing a total of 3011 participants, validated the inclusion criteria as shown in Fig. 1. Among them, two studies were added following expert consultation. These studies met the inclusion criteria but did not appear in the database search due to differences in indexing and terminology. Among the 34 studies, 38.24% were considered as a healthy elderly population, 32.35% with individuals subject to mild cognitive impairments, 17.65% with subjective cognitive disorder, and 14.71% with people with dementia. 88.24% of studies used an examination tool for the inclusion of participants (17 with the Mini-Mental State Examination [MMSE] and 16 with the Montreal Cognitive Assessment [MoCA]). We also emphasize that 29.41% of studies studied the psychological profile of participants with the Geriatric Depression Scale (GDS). Baseline characteristics of participants are presented in Table 1 while the intervention characteristics are in Table 2 and the post measures in Table 3.

Table 1 Population studies characteristics (Format: intervention / intervention // control / control with Male in %). Abbreviation: ADAS-JCOG, Alzheimer’s Disease Assessment Scale Japanese Cognitive Subscale; MMSE, Mini-Mental State Examination; HDSR, Hasegawa’s Dementia Scale Revised; BTEL, Buschke’s Test of Episodic Learning; MoCA, Montreal Cognitive Assessment; SVLT, Seoul Verbal Learning Test; DST, Digit Span Test; SWF, Semantic Word Fluency; PWF, Phonemic Word Fluency; GDS, Geriatric Depression Scale; 4-IADL, 4-item Instrumental Activities of Daily Living scale; TUG, Timed Up and Go test; HADS, Hospital Anxiety and Depression Scale; BDi, Beck Depression Inventory; ADL, Activities of Daily Living; ADAS-Cog, Alzheimer’s Disease Assessment Scale – Cognitive Subscale; CDR, Clinical Dementia Rating; BMI, Body Mass Index; SGDS, Short Geriatric Depression Scale; ABC, Activities-specific Balance Confidence scale; EQ-5D, EuroQol 5 Dimensions; MNA, Mini Nutritional Assessment; CRIq, Cognitive Reserve Index Questionnaire; LOTCA, Loewenstein Occupational Therapy Cognitive Assessment; MEC, Mini-Examen Cognoscitivo; QoL-G, Quality of Life – Geriatric version; QoL, Quality of Life; CIRS, Cumulative Illness Rating Scale; TMT, Trail Making Test; STAI-Y, State–Trait Anxiety Inventory, Form Y; MDPQ-16, Mobile Device Proficiency Questionnaire – 16 items; ToL, Tower of London; Ek-60F, Everyday Memory Checklist – 60 items (Form F); Mini-SEA, Mini–Social Cognition and Emotional Assessment; ISELS, Interpersonal Support Evaluation List – Short Form; SPMSQ, Short Portable Mental Status Questionnaire; GAD, Generalized Anxiety Disorder scale; CDR-SOB, Clinical Dementia Rating – Sum of Boxes; B-ADL, Bayer Activities of Daily Living Scale; K-IADL, Korean Instrumental Activities of Daily Living scale; MBI, Mild Behavioral Impairment; K-MoCA, Korean version of the Montreal Cognitive Assessment; K-MMSE, Korean version of the Mini-Mental State Examination; SVFT, Semantic Verbal Fluency Test; LSBQ, Language and Social Background Questionnaire; SUS, System Usability Scale; LM, Logical Memory; Cd, Coding; SS, Symbol Search; LFT, Letter Fluency Test; CFT, Complex Figure Test; SUBI, Subjective Well-Being Inventory; POMS, Profile of Mood States; IQCODE, Informant Questionnaire on Cognitive Decline in the Elderly; FAB, Frontal Assessment Battery; FCSRT, Free and Cued Selective Reminding Test; DSST, Digit Symbol Substitution Test; NPI, Neuropsychiatric Inventory; AI, Attention Index; AES, Apathy Evaluation Scale; BADL, Basic Activities of Daily Living; GDS-SF, Geriatric Depression Scale – Short Form
Full size table
Table 2 Characteristics of the intervention protocols
Full size table
Table 3 Cognitive tests used (at least in three studies) in post-intervention assessments. Abbreviation - ADAS-Cog: Alzheimer’s Disease Assessment Scale – Cognitive Subscale; DS: Digit Span; GDS: Geriatric Depression Scale; MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment; NPI: Neuropsychiatric Inventory; RAVLT: Rey Auditory Verbal Learning Test; TMT-A: Trail Making Test – Part A; TMT-B: Trail Making Test – Part B
Full size table

Quality and risk of bias

The quality and evidence in the included studies were assessed using the Rob2: Cochrane risk of bias tool for randomized trials [58]. We only kept for analysis studies without any high risk as presented in Fig. 2. The findings revealed that studies exhibited methodological bias, with 61.76% of them presenting some concerning features. The main reason for the problem of bias within the studies pertains to the lack of a protocol registered before the intervention (Domain 5). We also mention some concerns for 23.53% of studies regarding the randomization process not sufficiently explicitly presented (Domain 1).

Fig. 2
figure 2

Revised Cochrane risk-of-bias tool for randomized trials (RoB 2). Green corresponds to a low risk of bias and orange to some concerns

Full size image
Fig. 3
figure 3

Plot forest of Mini-Mental State Examination (MMSE) score in post intervention (with 6 studies involving 355 participants [\(n_{intervention}\) = 179, \(n_{control}\) = 176])

Full size image
Fig. 4
figure 4

Plot forest of Montreal Cognitive Assessment (MoCA) score in post intervention (with 5 studies involving 305 participants [\(n_{intervention}\) = 154, \(n_{control}\) = 151])

Full size image

Protocol divergence

Depending on the experiments, numerous protocol differences are utilized, and the measurement tools vary according to the studies. The most studied cognitive improvements are overall cognition with MMSE in 32.35% of cases, then inhibition and selective attention with Stroop in 20.59% of studies. Regarding the effect of interventions on mental health, depression is also measured in 11.76% of protocols, as shown in Table 3. A majority of studies do not use complementary treatments during the intervention. However, 41.18% conducted tests with one or more intervention protocols: 1 was carried out with participants undergoing pharmacological treatments, and 13 combined cognitive exercises with other non-pharmacological treatments (notably 9 with physical exercises and 5 with psycho-education). In terms of interfaces, 88.24% of studies employed tablets, while 14.71% of studies utilized smartphones and 8.82% employed touch tables. The duration of the training extended beyond 3 months in 41.18% of the studies, lasted between 1 and 3 months in 47.06%, and less than 1 month in 8.82% of the studies. The frequency of training sessions also varied, between less than two sessions per week in 14.71% of studies, while most of them required 3 to 6 sessions per week (82.35%). Otherwise, intervention was administered daily (2.94%). The prescribed duration per procedure ranged from 15 to 120 min, with most lasting between 15 and 45 min (64.71%). Training sessions were either supervised or unsupervised by a healthcare professional, with 44.12% of programs under controlled conditions, while others focused on autonomy or home care. The majority of studies employed a single control group; only 14.71% of them used more than one control group for analysis. There are also discrepancies regarding the type of control group used. 58.82% of studies used an active control group, 29.41% a waitlist group, and 23.53% a treatment as usual group.

Post-intervention results analysis

The thirty-four studies were included for data extraction and meta-analysis, taking into account the characteristics of the participants and the training programs. These studies utilized distinct cognitive and psychological state measures. Notably, the MMSE was employed as an outcome measure in 32.35% of studies, the MoCA in 17.65%, and the Geriatric Depression Scale (GDS) in 11.76%. Scores on cognition-related tests were also kept with the results of the Stroop in 20.59% of the studies, the n-back presented in 20.59%, the Digit Span in 17.65%, and the Trail Making Test in 17.65% of them.

General cognitive assessment

A total of six studies involving 355 participants (179 in the experimental group and 176 in the control groups) were utilized to analyze the post-intervention MMSE score. The improving results after intervention are statistically significant (mean difference: 1.76, 95% CI from 1.42 to 2.10, p \(\le \) 0.00001) with, however, a strong heterogeneity inherent to the variability of the studies (I2 = 91%), see Fig. 3. Besides, analysis of the results of studies utilizing the MoCA as an indicator of global cognition involves 5 studies with 305 participants (154 in interventions and 151 in control groups). The post-intervention improvement is statistically significant (mean difference: 1.99, 95% CI from 1.37 to 2.61, p \(\le \) 0.00001) as shown in Fig. 4.

Performance cognitive tasks

A total of 3 studies involving 457 participants (228 in the experimental group and 229 in the control groups) were utilized to analyze post-intervention scores on the digit span (DS), and the results were found to be statistically significant (mean difference: 0.61, 95% CI from 0.08 to 1.14, p \(\le \) 0.02). The results are presented in Fig. 5. 4 studies involving a total of 214 participants (115 in the experimental group and 99 in the control groups) were utilized to analyze post-intervention scores on the Trail Making Test and the results were found to be statistically significant (mean difference: −2.60, 95% CI from −3.56 to −1.65, p < 0.00001). The results are presented in Fig. 6. Concerning the other tests, the results were not statistically significant or analysis used different metrics not generalizable (i.e time, score, difference).

Fig. 5
figure 5

Plot forest of Digit Span (DS) score in post intervention (with 3 studies involving 457 participants [\(n_{intervention}\) = 228, \(n_{control}\) = 229])

Full size image
Fig. 6
figure 6

Plot forest of Trail Making Test (TMT) score in post intervention (with 4 studies involving 214 participants [\(n_{intervention}\) = 115, \(n_{control}\) = 99])

Full size image

Discussion

Main results

The results of this review about computerized cognitive training via touchscreen present four main conclusions. 1: A small effect of computerized cognitive training programs on global cognition measured by MMSE and MoCA, but limited by the sample size. This result does not allow us to conclude on the contribution of computerized programs to overall cognition [34]. 2: A statistically significant improvement in the Digit Span in post-intervention. This result shows a significant effect of interventions on short-term and working memory. 3: A statistically significant improvement in the Trail Making Test in post-intervention. This suggests an improvement in processing speed, attention, and flexibility after cognitive training on touchscreens. 4: A statistically significant improvement in the Geriatric Depression Scale, as presented in Fig. 7, suggested an impact of these interventions on depressive mood.

Fig. 7
figure 7

Plot forest of Geriatric Depression Scale (GDS) score in post intervention (with 4 studies involving 177 participants [\(n_{intervention}\) = 90, \(n_{control}\) = 87])

Full size image

Adherence in touchscreen cognitive trainings

A number of studies have already demonstrated that participants exhibit a lack of adherence to the cognitive training program in the long term. Thanks to the use of a more flexible solution characterized by touchscreens and gamified exercises, better results were expected. However, a substantial proportion of the examined studies report a significant percentage of dropouts. Several measurements were performed in the studies to estimate the adherence rate based on the individual characteristics of the participants. Baquero et al. [61] specify that they found no statistically significant relationship between a physical health variable and a person’s adherence to a computerized cognitive training program. Concerning the number of years of schooling, while some studies appear to emphasize a relationship between that factor and the public’s adherence to cognitive reeducation programs [61, 64] while others have not reported a link [65]. Future studies on computerized cognitive training ought to consider this difficulty to validate performance results.

Methodological caveats on touchscreen cognitive training

Quite a number of reviews have previously delineated methodological concerns pertaining to computerized cognitive training [41, 42]. It was hypothesized that implementing training on touchscreens had the capacity to facilitate the execution of strict protocols limiting attribution bias. Regretably, quite a number of biases were uncovered in this review. Difficulty in obtaining a sufficient sample size constitutes the primary stumbling blocks of conclusions in 61.76% of studies. The dematerialization of interventions does not appear to directly allow the membership of a significant group of individuals, nor the retention of a significant part of the initial sample. This renders the establishment of the requisite control groups unfeasible (waiting list and active control group). The use of well-designed control groups is an indispensable element of studies on computerized cognitive training inasmuch as placebo effects and participant expectations can influence observed outcomes [26, 66]. The presence of an active control group is important to ensure the valid interpretation of cognitive training. However, the presence of an active control group is not taken into account in a number of studies (29.41%), a methodological weakness emphasized in other reviews [59]. Especially since in a proportion of studies, the results attributable to the intervention program were not statistically significant when compared with the performance of an active control group [60, 63]. This suggests that the choice of the active control group must be carefully considered to highlight the mechanisms specific to computerized cognitive training programs compared to other numerical tasks [47]. However, when active control groups are not appropriately designed, it becomes difficult to establish intervention-specific effects against placebo effects [27]. Future studies should endeavor to obtain a more substantial sample of participants and emphasize the importance of implementing strict control groups. Furthermore, homogeneity or paucity of representativeness of the sample is frequently emphasized. In our sample, this feature characterized 44.12% of studies. For instance, gender proportionality is seldom attained despite some studies indicating a first-order relationship between gender and cognitive decline [67, 68].

Controlling external stimuli during interventions assumes paramount importance during unsupervised or independent interventions, and the diversity of program admissions is a weighty element to consider in the analysis of results. Initiating home studies confers an advantage in terms of access to patients whose physical condition or geolocation might otherwise preclude intervention in the center [61, 69]. Conversely, this administration of computerized cognitive training provides variability to program allocation models. Payne et al. [62] demonstrated that gains in processing speed in the home-based group reached 74.00% of those in the laboratory training. Finally, effects monitoring studies are not carried out systematically, despite the necessity to validate the long-term acquisition and transfer effects, given the ongoing debates on the effects of computerized cognitive training [41, 42]. About the results’ probable positive effect on “near transfer,” that is to say, on untrained tasks alike, disagreement remains present concerning the capacity for “far transfer” on more distant tasks requiring the same cognitive processes after a computerized cognitive training program [60, 70]. Nevertheless, few studies in our sample examined the transfer in their respective protocols. Furthermore, when transfer effects are studied, divergent results are frequently presented [59]. This may partly reflect, as for computerized cognitive training, the variability inherent to the transfer effects validation tasks. For example, Boujut et al. [80] employed virtual reality to examine the assessment of cognitive processes on a simulation of daily life tasks while Ballesteros et al. [59] used the same touchscreen platform with dedicated tasks included in the programs to examine the transfer. Among the studies considered, a third of them employed a multimodal training program. These programs refer to work suggesting that multimodal interventions may be more promising for cognitive improvement [71]. As pertinent as this approach might be, it does not justify the specific effectiveness of computerized cognitive training programs included in a global intervention.

Consideration of accessibility and usability

Motivation is a crucial factor of adherence to the program among seniors [72]. Furthermore, patients’ reduced engagement in cognitive training has been associated with the difficulty to perform exercises due to the symptoms of their condition [38]. Its importance is heightened by the context of remediation in cognitive sciences, as such programs typically require sustained and long-term participation to demonstrate efficiency [73]. This suggests that the user-friendliness and accessibility of computerized cognitive training programs are variables to consider to measure the effectiveness of treatments. Nonetheless, only few studies examine the accessibility of solutions before the prescription of the programs [80]. The present study focused specifically on computerized cognitive training programs presented on touchscreens. This follows the various works justifying the quality of these interfaces for individuals subject to cognitive disorders, as already highlighted by the first works in 1986 [18, 74, 75]. However, this is not sufficient, and computerized cognitive training ought to further consider personalization within programs [38]. This could be achieved with the integration of accessibility support tools, such as constant assistance or adaptation tools like Zoom [53]. This approach is not without introducing an additional complexity consisting of studying the multi-objective problem of minimizing accessibility difficulties in parallel with the quality of a remediation program. To facilitate the integration of patients in this approach, psycho-education approaches as well as opening to various highly personalized programs facilitating the participation of users should be considered [18, 47]. Another important point concerns the personalization and adaptation of computerized cognitive exercises on touchscreens according to the patient’s profile. Adapting the level of difficulty, the content, and user feedback can positively impact both engagement and the effectiveness of this type of intervention. Integrating adaptation algorithms and following a user-centered approach can improve the efficiency and long-term persistence of the exercises.

Strengths and limitations

The first strength of this study was to focus exclusively on RCTs applied on touchscreens attributions. Due to methodological limitations, we chose an analysis specifically focused on RCTs. The transition to computerized cognitive training on touchscreens enables the consideration of the implementation of cognitive training programs at home and the administration of the programs on interfaces accessible to people subject to cognitive disorders. However, the results of this review are limited by the number of included studies and the small average study sample size. Another limitation of the review is that publication bias was not addressed. Despite including 34 studies, the number of studies per outcome is insufficient to develop a funnel plot or Egger’s test. Moreover, even if we only retained the RCTs, the methodologies applied are variable, in addition to encountering problems related to protocol difficulties. A pronounced heterogeneity was also observed among the results between experiments. This postulate prevents the claimed results from being strictly validated. Additionally, our review included language restrictions and a limited time frame (9 years) to narrow the scope to more modern and improved accessibility interfaces, ignoring consumer touchscreens that have appeared since 2010. For these reasons, it is difficult to draw conclusions on the contribution of computerized cognitive training on a touchscreen even if the preliminary results seem to justify an interest in prevention and cognitive decline. Overall, the certainty of the evidence presenting computerized cognitive training on touchscreens as a tool for cognitive remediation is low to moderate. This result reflects the small sample sizes, identified concerns about the risk of bias, and the heterogeneity of the protocols and outcomes measured. Although 19 studies reported follow-up outcomes, these were short-term results, ranging from 1 week to 3 months. Therefore, the long-term effectiveness of cognitive interventions cannot be established. Future research incorporating long-term follow-up is needed.

Conclusion

The results of this review demonstrate that computerized cognitive training administered on a touchscreen appears to be feasible and adhered to by a large proportion of seniors. Their utilization appears to be beneficial for a consequent percentage of that population. However, many of the issues presented in reviews targeting cognitive training are also persisting in this context. The effects on overall cognition and on specific cognitive processes such as episodic memory and working memory are visible and significant. At the same time, notable positive results were uncovered for short-term memory and attention. However, the methodological difficulties and the variability of practices prevent us from strictly concluding on the effectiveness of playful cognitive training on touchscreens. Conversely, subsequent work should explore the correlation between accessibility of program distribution environments and the outcomes on performance attributable to training, as well as the unique characteristics of the patient’s initial conditions.