Does Dosage Drive Durability? A Longitudinal Study on Digital Cognitive Training for Sustained Functional Independence Bruno Costa Poltronieri, Cíntia Monteiro Carvalho, Karin Reuwsaat, Maria Eduarda Alves Reis, Ana Carolina Madeira, Thayná de Carvalho Freire, Rogerio Panizzutti Abstract ABSTRACTBackgroundThis study aimed to investigate whether a higher cumulative dose of digital cognitive training leads to greater and more sustained improvements in cognition and functionality among older adults with Mild Neurocognitive Disorder.MethodsWe recruited participants aged 60 or older with mild neurocognitive disorder to a double‐blinded, randomized, stepped wedge clinical trial of digital cognitive training (DCT) compared to an active control group of commercial computer games. Participants were evaluated for functionality and cognition before, after 10 h and 20 h of intervention and in the follow‐up.ResultsSixty‐seven participants completed DCT and were evaluated. The group that received 20 h of DCT showed significant improvements in functionality and satisfaction, as well as gains in processing speed, verbal fluency, and inhibitory control. The 10 h group also showed improvements in specific areas such as processing speed and recognition. In the 20 h group, the benefits were maintained at 3‐ and 6‐month follow‐ups, especially in functionality.DiscussionThis study suggests that a higher cumulative dose of DCT may lead to more sustained improvements in functionality performance among older adults with mild neurocognitive disorder. Findings highlight the need for further research to determine optimal training dosage for long‐term improvements in cognition and functionality.Trial RegistrationThe studies reported in the manuscript were pre‐registered at ClinicalTrials.gov under the identifier NCT03911765 Randomized trial of 67 older adults comparing 10 h vs. 20 h cumulative digital cognitive training doses. Both show immediate gains but only the higher dose sustains long‐term functional benefits.[GGI-26-0-g002.webp] Introduction Global population aging presents major health and social challenges, especially regarding cognitive decline from normal aging to dementia []. Mild neurocognitive disorder is defined as an intermediate stage between normal cognitive aging and dementia, involving deficits in specific cognitive domains without impairment of daily activities [, , ]. Although mild neurocognitive disorder does not significantly impair overall functionality, older adults with this condition may experience difficulties performing instrumental activities of daily living []. Research has shown a strong association between executive functions and daily functioning [, ]. These findings highlight the importance of developing and investing in interventions, such as digital cognitive training (DCT), that support and enhance executive functions. DCT has been increasingly explored as an intervention that can mitigate cognitive decline in older adults, such as mild neurocognitive disorder. DCT encompasses structured practice with standardized and cognitively demanding tasks [] and provides various benefits, such as engaging visual interfaces, efficient and scalable implementation, and the capacity to dynamically adjust training content and difficulty based on individual performance [, ]. Meta‐analysis on DCT in older adults with mild neurocognitive disorder has reported effects ranging from small to moderate for global cognition, working memory, memory, and learning [, ]. Despite these promising results, some challenges remain unclear regarding DCT, including the optimal cumulative dose and the duration of cognitive effects in older adults with mild neurocognitive disorder. Evidence suggests that sessions shorter than 30 min are ineffective [], while around 50–55 min, six times per week, may yield better outcomes []. However, it is still uncertain whether higher cumulative doses produce greater benefits or if the effects persist long term, as most studies assess only immediate outcomes [, ]. Some research, however, shows sustained benefits, such as global cognition improvement 3 months post‐training [] and maintained episodic memory gains 6 months later []. Yet, neither study examined the long‐term impact of different cumulative doses of DCT in older adults with mild neurocognitive disorder. Recently, our group conducted a stepped wedge randomized clinical trial in which we explored the effects of DCT exercises primarily focused on enhancing executive functions on functionality and cognition in older adults with mild neurocognitive disorder []. To assess functionality, we employed the Canadian Occupational Performance Measure (COPM), an individualized tool widely used in rehabilitation settings across several countries, designed to identify difficulties in activities of daily living and provide a comprehensive understanding of individual challenges. Participants who underwent 10 h of DCT showed significant improvements in COPM scores and learning compared to those who engaged in leisure games []. Building on these findings, the present study examines the impact of 10 or 20 h of DCT on functionality and cognition, immediately after training and 3 and 6 months after the end of training. We pose the following questions: (1) Are there significant differences in the effects of 10 or 20 h of DCT on functionality and cognition after training? (2) Are there significant differences in the effects of 10 h or 20 h of DCT on functionality and cognition at 3 and 6‐month follow‐ups? Methods Participants Participants were recruited through professional referrals and social media advertisements and contacted the research team via email or text message. Eligibility criteria for inclusion in the study were: age ≥ 60 years; normal or corrected vision and hearing; access to the internet; and diagnosis of Mild Neurocognitive Disorder based on DSM‐5 criteria. Exclusion criteria included: a diagnosis of dementia; dependence on activities of daily living; and major medical conditions that precluded study participation. All participants underwent a thorough evaluation, which included clinical history, cognitive and functional assessments, and evaluations of anxiety and depression symptoms. The clinical diagnosis was established in accordance with DSM‐5 criteria and confirmed by a board‐certified psychiatrist (R.P.). Participants were assessed during the study period from July 2020 to January 2025. Before enrollment, all participants provided written informed consent after receiving a detailed explanation of the study procedures. The research was approved by the Ethics Committee of the Federal University of Rio de Janeiro—Institute of Psychiatry (4.315.008) and pre‐registered at ClinicalTrials.gov (NCT03911765). Study Design The study was a randomized, double‐blinded, stepped wedge clinical trial []. Participants were randomized by an independent researcher using a hierarchical stratification by age, education, MoCA score, gender, Technology activities of daily living questionnaire (TADLQ), and randomly assigned to one of the groups. Initially, only one group performed 10 h of DCT and the other group played 10 h of commercial games. Then, both groups performed an additional 10 h of DCT. Thus, one group performed a total of 10 h of DCT and the other performed 20 h of DCT. We performed the follow‐up assessments three and 6 months after the end of DCT. All recruited participants (n = 67) were assessed for eligibility, and those who met the inclusion criteria underwent cognitive and functional assessments before starting the training, immediately after 10 h and 20 h of training, three and 6 months after the training for follow‐up. The assessment battery was applied by a team of clinicians composed of occupational therapists and neuropsychologists. Participants did not receive any financial incentive to participate in the study. Participants trained at home either on their own or on a study‐provided laptop or tablet. They were asked to train as much as possible, 1 h per Session 2 twice a week. Subjects in both groups were remotely monitored via video call by a designated research assistant, who provided support with issues such as technical issues or adherence to the training protocol. Measurements An assessment battery was administered to measure participants' cognition and functionality. Regarding cognitive domains, we investigated verbal fluency, memory (including long‐term memory, short‐term memory, and recognition), learning over trials, processing speed, cognitive flexibility, and global cognition. To this end, the following instruments were used: the Montreal Cognitive Assessment (MoCA) [, ] to assess global cognition; the Verbal Fluency Test to assess verbal fluency []; the Rey's Auditory‐Verbal Learning Test (RAVLT) [, ] to evaluate long‐term memory (list A7—number of words recalled after a 20‐ to 30‐min delay), short‐term memory (list A6—number of words recalled after the interference list, with scores ranging from 0 to 15), recognition (list A and non‐list A words correctly identified), and learning over trials (sum of words recalled across five trials minus five times the first trial, with scores ranging from 0 to 75); and the Stroop Test [] to measure processing speed and inhibitory control, referring to card 1 and card 3, respectively. To evaluate functionality, we applied the Canadian Occupational Performance Measure (COPM) [] and the Technology Activities of Daily Living Questionnaire (TADLQ) []. The COPM is a semi‐structured, client‐centered interview designed to assess participants' self‐perceived occupational performance and satisfaction in relation to specific individualized functional challenges [], whereas the TADLQ is a structured questionnaire that evaluates the functionality of older adults to perform daily activities, including a specific subscale for technology use []. To assess neuropsychiatric symptoms, we applied the Geriatric Depression Scale 15 (GDS) [] and Geriatric Anxiety Inventory (GAI) []. Cognitive Training Exercises The current study was based on the hypothesis that training with digital exercises focused on executive functions could improve the functionality of the participants []. Thus, we designed the digital neuroscience‐informed cognitive training using a series of exercises from the BrainHQ platform (www.brainhq.com, PositScience Inc.) focused on improving executive functions. The exercises were: Divided Attention, Eye For Detail, Target Tracker, Juggle Factor, Double Decision, Right Turn, Mental Map, and Optic Flow. The 10 h cognitive training group began the first phase of the study by engaging in computer‐based leisure games, including The Birds Game, Let's Clean Up, Bubble Poke, Word Search, Memory Game, Puzzle, Dominoes, Smarty Bubbles, and Spot the Difference. After completing 10 h of these leisure games, participants proceeded to a second phase consisting of 10 h of DCT. The 20 h cognitive training group completed the cognitive exercises throughout the study, practicing each exercise for 15 min per session until completing 1 h of DCT per session. Full details about the group conditions are provided in Carvalho et al. []. Statistical Analysis We verified data normality using the Shapiro–Wilk test. Winsorized means were calculated for verbal fluency, GDS, TADLQ, processing speed, inhibitory control, and learning over trials and constituted 10% of the total data. Measures were normally distributed after winsorizing of outlying values. We used t‐tests and chi‐square tests to compare sociodemographic characteristics, functionality, cognition, and neuropsychiatric symptoms between groups at baseline. We used a mixed‐model analysis of covariance (ANCOVA) to determine the effects of DCT, using baseline scores and education as covariates. Post hoc analyses used one‐way repeated‐measures analysis of variance (RM‐ANOVA), with Sidak corrections to adjust for multiple comparisons. Time (pre‐training, post‐training, 3‐month follow‐up, and 6‐month follow‐up) was treated as the intra‐group factor, and training dose as the inter‐group factor. Spearman's rho correlation was used to assess the association between cognition and functionality. Only participants who completed all study phases were included in the analysis. We used SPSS version 23 for statistical analyses. Cohen's d values were derived from eta squared. Results Participants We assessed 138 participants and excluded 51 because of ineligibility or lack of interest in the study. Eighty‐seven participants were randomly assigned to one of the two groups: 10 h of DCT (n = 45) or 20 h of DCT (n = 42). Of the total number of randomized participants, 67 completed the protocol of DCT, while 41 were evaluated 3 months post‐training of DCT, and 31 were evaluated 6 months post‐training. Post‐training attrition rates were 24.4% in the 10 h DCT group and 21.4% in the 20 h DCT group. At the 3‐month follow‐up, attrition increased to 38.3% in the 10 h group and 39.3% in the 20 h group. By the 6‐month follow‐up, attrition rates rose further to 53% for the 10 h group and 54.5% for the 20 h group. Participants were predominantly composed of women with 14 years of education (SD = 5.2) and a mean age of 74.9 (SD = 7.2). There were no significant sociodemographic and clinical differences between the two randomized groups (Table ). Changes in Functionality and Cognition Post‐Training Mixed‐model ANCOVA using the baseline scores and education as covariates revealed a global effect over time post‐training in COPM performance (F (1, 54) = 6.18, p = 0.016, d = 0.67) and in COPM satisfaction (F (1, 51) = 9.21, p = 0.004, d = 0.85) with no significant group‐by‐time interaction for both groups. Post hoc one‐way RM‐ANOVA for the two groups separately revealed significant improvement over time in COPM performance to 20 h of DCT group (F (1, 22) = 14.5, p = 0.001, d = 1.62) and also to COPM satisfaction (F (1, 24) = 12.05, p = 0.002, d = 1.41). There was no change in COPM performance and COPM satisfaction for 10 h DCT group (Image 1). Additionally, there was no effect on the TADLQ measure of functionality for 20 h of DCT group, without effect over time for both groups (Table ). The analysis of individual cognitive domains, controlling for multiple comparisons, showed a significant effect of time for both groups over 20 h of training for verbal fluency, learning over trial, recognition, processing speed, inhibitory control and short term memory (Table ). Post hoc analyses demonstrated improvements over time in specific domains for the 10 h DCT group, including processing speed (F (1, 31) = 14.81, p = 0.001, d = 1.38), recognition (F (1, 25) = 33.44, p < 0.001, d = 2.31), learning over trials (F (1, 25) = 9.74, p = 0.004, d = 1.24), and inhibitory control (F (1, 31) = 10.01, p = 0.003, d = 1.13). Similarly, the 20 h of DCT group showed significant improvements over time in processing speed (F (1, 29) = 3.31, p = 0.04, d = 0.67), recognition (F (1, 26) = 15.23, p = 0.001, d = 1.52), learning over trials (F (1,26) = 14.68, p = 0.001, d = 1.50), and verbal fluency (F (1, 30) = 4.97, p = 0.03, d = 0.81). However, no significant effect on short‐term memory was observed for either group following the post hoc analyses. Finally, we conducted a Pearson correlation to identify which changes in cognitive domains were associated with improvement on the COPM. A significant correlation was observed between this variable and verbal fluency (r = 0.39; p = 0.02) in the group that underwent 20 h of DCT. Changes in Functionality and Cognition in the Follow‐Up In order to examine the durability of the effects, we performed a follow‐up assessment after 3 months (Table ) and 6 months (Table ) of the end of training sessions. The follow‐up assessment revealed a main effect of time in COPM performance for the 20 h of DCT group after 3 months (F (1, 14) = 14.0, p = 0.001, d = 2.37) with a significant group‐by‐time interaction (F (1, 33) = 14.11, p = 0.001, d = 1.03). These results persisted until 6 months after the end of the training program, with improvements over time (F (1, 25) = 29.07, p < 0.001, d = 1.32) with a significant group‐by‐time interaction (F (1, 25) = 9.03, p = 0.006, d = 1.20) for the 20 h of DCT group. With regard to COPM Satisfaction, the post hoc analyses showed improvements in COPM satisfaction for 20 h of DCT group (F (1, 12) = 18.35, p < 0.001, d = 2.47) without effects for 10 h of DCT group. However, in the assessment performed after 6 months, the COPM satisfaction improved significantly over time (F (1, 10) = 10.92, p = 0.008, d = 2.09) for 20 h of DCT group. There was no change in COPM Performance and COPM satisfaction in 10 h of DCT group in follow‐up (Figure ). When analyzing individual cognitive domains, no group‐by‐time interaction was observed in follow‐up. However, there was a significant effect of time in processing speed for 10 h of DCT (F (1, 17) = 6.200, p = 0.02, d = 1.20) and 20 h of DCT group (F (1, 18) = 20.70, p = 0.001, d = 2.14). These improvements in processing speed persisted at the 6‐month follow‐up assessment for both the 10 h DCT group (F (1, 12) = 9.01, p = 0.01, d = 1.73) and the 20 h of DCT group (F (1, 12) = 6.64, p = 0.02, d = 1.48). Regarding inhibitory control at the 3‐month follow‐up, the 10 h DCT group showed a trend toward improvement in inhibitory control (F (1,17) = 3.68, p = 0.07, d = 0.93), while the 20 h DCT group demonstrated a significant improvement over time (F (1, 18) = 5.92, p = 0.02, d = 1.14). At the 6‐month follow‐up, both groups exhibited significant improvements: the 10 h DCT group (F (1,12) = 5.09, p = 0.04, d = 1.30) and the 20 h DCT group (F (1,12) = 14.17, p = 0.003, d = 2.17). For recognition, the 10 h DCT group showed significant improvement at the 3‐month follow‐up (F (1,18) = 35.74, p < 0.001, d = 2.81), while the 20 h DCT group did not (F (1, 17) = 0.75, p = 0.39, d = 0.04). At the 6‐month follow‐up, the 10 h DCT group maintained significant improvement (F (1, 12) = 7.50, p = 0.01, d = 1.58), whereas the 20 h DCT group did not reach significance (F (1, 12) = 2.44, p = 0.14, d = 0.90). Finally regarding learning over trials, at the 3‐month follow‐up, both the 10 h DCT group (F (1, 14) = 15.07, p = 0.002, d = 2.07) and the 20 h of DCT group (F (1, 13) = 5.44, p = 0.03, d = 1.29) demonstrated significant improvements; however, at 6‐month follow‐up there are no significant effects over time for both groups. Discussion This study investigated the dose‐cumulative effects of 10 and 20 h of DCT on functionality and cognition in older adults with mild neurocognitive disorder, and found that both groups improved cognitively after training; however, only the 20 h of DCT group maintained significant functional gains up to 6 months later, supporting our initial hypothesis [] that a higher training dose leads to more durable cognitive and functional benefits. Our findings are consistent with those reported in the ACTIVE study [], where, after 10 years, participants who performed DCT (in three intervention groups) reported fewer difficulties in performing instrumental activities of daily living compared to those in the control group. However, it is important to note that the result in the ACTIVE study was only observed between years five and ten, which differs from our study due to its shorter follow‐up period. Unlike our findings, other studies that investigated functionality in older adults with mild neurocognitive disorder by scale or questionnaire to measure daily functioning following cognitive remediation, which includes DCT, virtual reality interventions, or neuropsychological rehabilitation, did not report significant changes in this variable [, , ]. There are several possible explanations for the sustained effects of training on self‐reported daily functioning, as measured by COPM performance and satisfaction. Our DCT protocol primarily targeted executive function, a broad cognitive domain encompassing initiation, sequencing, monitoring, inhibition, set‐shifting, planning, problem‐solving [], verbal fluency, and other components []. Many of these subdomains are moderately associated with performance in daily functioning []. Our previous study [] found that improvements in COPM performance were associated with gains in processing speed and verbal fluency—both of which showed significant long‐term effects in the 20 h of DCT group. In the present study, however, we found an association only with change in verbal fluency. The maintenance of COPM‐related effects may also be explained by the behavioral changes induced by the cognitive training. The cognitive improvements may have encouraged participants to re‐engage in activities they were previously interested in or motivated them to explore new occupations []. Regarding the instrument employed to evaluate functionality, studies evaluating the validity of the COPM in different contexts have shown that this instrument demonstrates consistent content validity, construct validity, and feasibility [], suggesting that it is a useful tool for assessing daily function. Furthermore, it addresses self‐perceived problems that may not be captured by other standardized measures []. The cognitive improvements observed following training occurred irrespective of whether participants received 10 or 20 h of intervention. This finding is in line with prior research indicating that both short‐ and long‐duration training protocols tend to produce comparable outcomes in terms of near transfer effects; however, extended protocols may enhance certain aspects of far transfer effects []. A study published by [], which investigated cognitive outcomes before, during, and after multi‐domain DCT, revealed a marked therapeutic response, with notable improvements even after a limited number of training sessions. Similarly, previous research has shown acute brain changes following 8 weeks of cognitive training [], although evidence for the long‐term persistence of these changes remains limited. Despite these encouraging findings, the relationship between dose‐cumulative training and cognitive outcomes remains a subject of debate in the literature. A meta‐analysis reported that a lower total number of sessions was associated with more significant long‐term effects [], which contrasts with another study involving individuals with schizophrenia, suggesting that higher doses of DCT may be necessary to promote lasting cognitive gains []. This discrepancy reinforces the notion that the optimal dosage of DCT is not yet well‐established, particularly regarding its long‐term impact []. Considering that cognitive gains do not appear to follow a linear trajectory, it is essential for future research to further investigate the role of dose‐cumulative training and explore how maintenance training may influence outcomes over time. In addition to behavioral outcomes, numerous studies have investigated the neurobiological mechanisms underlying the cognitive improvements associated with cognitive training. DCT has the potential to promote neuroplasticity by engaging the brain in targeted cognitive activities that enhance its functional capacity [], a process that may have occurred among the participants in the present study. The effect of neuroplasticity induced by DCT was observed in studies using the diffusion tensor imaging technique. These studies have demonstrated that skills training over several weeks promotes microstructural changes of limbic system structures, reflecting changes in the rearrangement of neural tissue [, ]. Additionally, memory training has been linked to increased activation and connectivity in various brain regions, such as the frontal, temporal, and occipital lobes, as well as the hippocampus, in both cognitively healthy older adults and those with mild neurocognitive disorder. However, the exact extent of these changes in individuals is still not fully understood [, , , ]. As far as we know, this is one of the few studies to examine the cumulative dose of DCT using a stepped‐wedge clinical trial design in older adults with mild neurocognitive disorder, with follow‐up assessments conducted in the absence of additional training. However, this study has several limitations: (1) As in most follow‐up studies, we experienced a significant attrition rate at the 6‐month assessment, which reduced the statistical power of our sample during that phase; (2) Since the majority of this study was conducted during the COVID‐19 pandemic, most participants recruited were older adults with access to technology and the internet. This group is not representative of the broader Brazilian population. To mitigate this limitation, we provided tablets to participants residing in Rio de Janeiro and offered support via videoconference; (3) Although our functionality measures are sensitive and cover diverse constructs, they rely on self‐report and self‐perception. It is important for future research to also incorporate ecologically valid and performance‐based assessment tools, which may offer better insight into the transfer of cognitive training gains to everyday functioning in older adults exposed to varying doses of DCT; (4) Finally, it is important to highlight that due to the high attrition rate during follow‐up, it was not feasible to perform an intention to treat analysis, although the authors acknowledge the importance of such analyses in randomized clinical trials. Although definitive conclusions cannot be drawn, it is possible that the cumulative dose of cognitive training contributed to the sustained effects observed, particularly in the COPM outcomes in our study. In addition, while the 20‐h dose was associated with sustained functional gains, it is important to acknowledge potential confounding factors that may have influenced these outcomes during the 6‐month follow‐up. Variables such as baseline cognitive reserve, levels of physical exercise, and participation in social or cognitive activities were not strictly controlled and could have played a role in maintaining independence. Although the study design aimed to isolate the effects of DCT, these external factors should be considered when interpreting the durability of functional improvements in older adults with Mild Neurocognitive Disorder. Future research should further explore the ideal training dosage, so that clinicians can implement DCT programs with greater precision and effectiveness for older adults with mild neurocognitive disorder. Ethics Statement The research was approved by the Ethics Committee of the Federal University of Rio de Janeiro—Institute of Psychiatry (4.315.008). Consent All participants gave informed written consent prior to study inclusion for their information to be used for research, and were treated in accordance with the declaration of Helsinki.