Both studies were approved by the Medical University of South Carolina’s (MUSC) Institutional Review Board and were pre-registered in ClinicalTrials.gov (aMCI: NCT04503096; and psMCI: NCT04655963; IRB Pro00100536 and Pro00083136, respectively). All participants provided written informed consent and were informed of their right to withdraw from the study at any time without penalty. Participant privacy and confidentiality were strictly maintained in accordance with IRB requirements, and no personally identifiable information is reported in this manuscript or supplementary materials. Participants were offered compensation at the rate of US $4.50 per completed 20-minute session per day, equating to up to US $126 total. Efforts to reduce potential sources of bias in these open-label trials included (1) minimizing selection bias by using rigorous and predefined eligibility criteria applied consistently to all participants; (2) minimizing confounding factors by constraining disease severity to MCI and excluding comorbidities likely to impact the study outcomes; and (3) avoiding reporting bias through trial pre-registration and a transparent and complete description of the study procedures, outcome measures, hypotheses, planned analyses, and results. The primary results from the aMCI trial focused on safety, feasibility, and acceptability of iTBS are reported in a study by Aghamoosa et al [26]. The description of the BrainHQ methods adheres to the Template for Intervention Description and Replication (TIDieR) checklist [27], as described below and summarized in Table S1 in Multimedia Appendix 1.

Differences in demographic and clinical characteristics between groups were examined using Wilcoxon rank-sum tests and chi-squared or Fisher exact tests, as appropriate. We present descriptive statistics to summarize adherence, engagement, and acceptability for each group. Wilcoxon rank-sum tests were used to compare the total number of minutes of engagement with the BrainHQ exercises between the aMCI and psMCI groups during the supervised treatment phase. During the unsupervised follow-up phase, a rank-based nonparametric alternative to a repeated-measures ANOVA was performed using the nparLD 2.2 package in R (4.4.2), which produces ANOVA-type statistics (ATS) [34]. This analysis tested whether there was a difference in the number of minutes of engagement in BrainHQ by the MCI group across all weeks as well as whether engagement changed across the weeks by group (ie, a group-by-week interaction).

Demographic and cognitive characteristics are summarized in Table 1. Consistent with the differing age range criteria in the 2 studies, participants in the aMCI group were older than those in the psMCI group. The groups did not differ significantly by sex, and both groups were predominantly non-Hispanic White. Consistent with their distinct underlying etiologies, the 2 MCI groups demonstrated differences in cognitive functioning (ie, poorer memory and semantic fluency in the aMCI group and poorer processing speed in the psMCI group).

Retention was high in both studies. One treatment initiator withdrew from the aMCI group (ie, 21/22, 95.5% completed all study procedures). All participants in the psMCI group who initiated treatment completed study procedures (ie, 14/14, 100%).

Adherence to BrainHQ during the in-laboratory treatment phase was similar in the 2 groups, with both groups completing approximately half of the prescribed 210 minutes: aMCI group median: 51.1% (IQR 43.6%‐57.6%; range 12.9%‐79.5%) of minutes and psMCI group median: 52.9% (IQR 44.4%‐65.3%; range 44.1%‐78.4%; see see Table 2 for the median number of minutes). No participant in either group completed 210 minutes during the treatment phase. Ecological adherence to the target 560 minutes of BrainHQ during the 4-week follow-up phase was higher in the aMCI group (median 80.2%, IQR 29.9%‐103.7% of minutes; range: 0%‐172.4%) than in the psMCI group (median 25.3%, IQR 0%‐85.3% of minutes; range: 0%‐192.4%). The percentage of participants who completed the target 560 minutes over the 4 weeks was 33.3% in the aMCI group and 21.4% in the psMCI group. The percentage of participants who met weekly goals (ie, target 140 min/wk) was variable in the aMCI group: week 1, 2 (10%) participants; week 2, 12 (57%) participants; week 3, 5 (24%) participants; and week 4, 7 (33%) participants. The percentage of psMCI participants who completed the target 140 minutes each week was lower and relatively stable: week 1, 3 (21%) participants; week 2, 5 (36%) participants; week 3, 3 (21%) participants; and week 4, 2 (14%) participants. In both groups, no participant (ie, 0%) completed the target 28 sessions of daily BrainHQ. The aMCI group median number of sessions was 16 (IQR 9‐21), and the median for the psMCI group was 4 (IQR 1‐14).

The median and IQR minutes of BrainHQ exercises completed on each treatment day and during the 4-week follow-up period are included in Table 2 and depicted in Figure 1. The mean minutes spent engaging in exercise domain on each treatment day and during each week of the 4-week follow-up period are depicted in Figure 2.

Medians and interquartile ranges for the items from the study-specific experience questionnaires are reported in Table 4 for the aMCI group and Table 5 for the psMCI group. Of note, although not all participants engaged with BrainHQ during the follow-up phase, all participants provided ratings of their experience immediately posttreatment. Thus, in Table 4, we present ratings separately for all participants post-treatment, all participants post-study, and the subset of participants who engaged in any BrainHQ during follow-up.

Participants in the aMCI group, on average, rated the BrainHQ exercises as making their cognitive functioning and mood “somewhat better” at both post-treatment and 4-week follow-up. The aMCI participants generally rated their experience with BrainHQ positively at both time points; they found the exercises fun, appropriately challenging, not tiresome or frustrating, and they indicated that they tried their best. Finally, aMCI participants rated their experience with the BrainHQ platform as positive, finding it easy to navigate and access and adequately explained, understanding what to do, and being generally satisfied with the feedback from the BrainHQ platform at both time points.

Participants in the psMCI group generally rated the BrainHQ exercises as “somewhat” to “very much so” helpful in improving mental abilities and everyday functioning both post-treatment and at 4-week follow-up. psMCI participants also rated their experience with BrainHQ positively in terms of their motivation, trying their best, and enjoyment, and overall did not find BrainHQ exercises time-consuming, difficult or burdensome, or boring at both time points. On average, psMCI participants rated the platform positively, finding that it was easy to understand and that they had interest in continuing, interest in learning more about it or participating in similar studies, and that they would recommend the platform. Ratings of BrainHQ were similar posttreatment and post 4-week follow-up, despite somewhat limited engagement with BrainHQ in the follow-up period.

These 2 pilot studies provide preliminary support for the feasibility and acceptability of interleaving CCT with accelerated iTBS for people with MCI, given that there was comparably high engagement in the BrainHQ exercises during the in-laboratory treatment phase across both groups. This was consistent with our hypothesis, given the structured nature of administration after each accelerated iTBS session. Although, on average, participants completed only approximately half of the possible 210 training minutes, this finding is somewhat expected. The 10‐ to 15-minute inter-session intervals used for delivering CCT also had to include time spent navigating the platform, transitioning in and out of BrainHQ from iTBS sessions, and taking breaks (eg, resting in between sessions and using the restroom). Future studies using intermittent CCT will need to account for these timing factors when planning training time.

The groups differentially engaged with BrainHQ during the 4-week unsupervised follow-up phase. The engagement rate increased in the aMCI group over the 4 weeks, with most participants independently engaging with BrainHQ in the follow-up phase. Engagement in the psMCI group was numerically lower, although not statistically significant, and decreased over time, with the majority of participants not engaging with BrainHQ in follow-up weeks 3 and 4. Generally, this indicates that ecological adherence to CCT following iTBS may be feasible for many persons with MCI, although some patients may require more support (eg, regular reminders, in clinic, with care partner) to reach CCT engagement goals following iTBS. Notably, while several in the aMCI group and some in the psMCI group met the prescribed number of minutes in the follow-up phase, no participant in either group completed daily BrainHQ practice (ie, all 28 d). This may suggest that a minute-based target may be more attainable than a day-based one or that daily training may not be feasible in these populations. For both groups, the most commonly completed exercises were in the domains of attention and brain speed in both the treatment and follow-up phases. During the follow-up phase, the psMCI group engaged in somewhat more memory and intelligence exercises than those with aMCI; however, differences in exercise choice are difficult to interpret, given the lower engagement in this group. Finally, despite this differential engagement, both groups rated the acceptability of the BrainHQ exercises positively.

One of the most salient findings was that during the self-administered follow-up phase, the aMCI group, on average, completed closer to the recommended BrainHQ engagement time of 140 minutes per week, whereas the psMCI group, on average, did not complete the prescribed number of minutes, and engagement decreased over time. Likely contributors to this observed difference in ecological adherence are differences in study design. Notably, the aMCI study provided weekly email reminders to complete the daily CCT, whereas the psMCI study did not. These findings suggest that in cognitively impaired samples, the use of regular reminders and other remote supports may be necessary to support completion of CCT outside of the laboratory setting. Another potential factor is that the aMCI group was offered the à la carte option during follow-up, whereas the psMCI group was instructed to continue using the personal trainer. Allowing aMCI participants to self-select exercises (via à la carte) may have invoked greater autonomy over their training experience, leading to increased adherence.

It is also possible that disease-specific factors could have contributed to discrepant engagement in CCT during the self-administered follow-up phase. The aMCI group may have had greater motivation, perceived benefit, or expectancy of benefit from CCT, given that these participants knew that, given the neurodegenerative nature of AD, they would likely progressively decline. In comparison, the psMCI group, while still experiencing objective impairments, would have already experienced significant spontaneous recovery from their initial deficits due to the stroke, with a prognosis of stability of functioning if other vascular events are avoided. The type of cognitive deficits experienced by the 2 groups may have also contributed. As expected, the aMCI group had primarily amnestic impairment along with set-shifting and semantic fluency difficulties. In comparison, the psMCI group performed significantly worse on a test that involves cognitive processing speed and visuo-motor organization (ie, coding), which are processes required for completion of all BrainHQ exercises. The psMCI group also had more variable deficits, which is expected given that cognitive sequelae poststroke are heterogeneous. Additionally, while significant deficits that would affect the ability to complete cognitive tasks were exclusionary, mild visual and motor deficits are not uncommon post-stroke [35]. Most BrainHQ exercises, and particularly those in the domains of attention and processing speed, which were most often completed in this study, are visual in nature and require a fast response with the mouse or a finger. Therefore, the psMCI compared to the aMCI group may have found these exercises more challenging as a function of sensorimotor impairments that compounded the cognitive demands of the exercises, thus impacting their engagement.

This is the first study to our knowledge to elicit feedback on participants’ experience with the BrainHQ platform interleaved with accelerated iTBS. Although the study-specific surveys differed, limiting direct comparisons, results from both surveys generally indicated that participants in both groups found it relevant and potentially beneficial to their functioning, enjoyable and not excessively burdensome, and feasible for use during both treatment and post-treatment periods. The psMCI group rated BrainHQ favorably despite limited engagement in the follow-up post-treatment phase, so this questionnaire did not aid in identifying why engagement was lower compared to that in the aMCI group. Future research using qualitative and mixed methods may elucidate the factors driving differential engagement, which could inform CCT delivery in these populations. Despite the small-scale nature of these pilot studies, our findings highlight the importance of individualizing interventions to the specific population and underscore the need to avoid treating all cases of MCI as homogeneous syndromes when planning treatment approaches.

Prior work investigating combined neuromodulation (eg, rTMS or transcranial direct current stimulation) and cognitive training has provided some additional insights into these combined interventions in the context of AD [36-38] and poststroke [39]. Although a growing literature supports the potential additive benefit of combining noninvasive brain stimulation (including rTMS) with cognitive training (including CCT; [40,41]), notably missing from the literature are studies refining methodologies for such combined interventions. Systematic reviews of the CCT literature demonstrate that CCT methodology is quite heterogeneous in terms of platform, training exercises, dose, and setting of administration [12,42]. While there is some evidence of the benefit of CCT to the same cognitive construct, there is limited support of benefits to different cognitive constructs and everyday activities [16]. However, many of these studies may not have been dosed for efficacy and used platforms with limited empirical support [21]. CCT was not explicitly dosed for efficacy in the current feasibility studies; however, a “gold standard” dose to elicit change has not been established [21]. Overall, the mixed literature and varied methodology suggest that there needs to be more research to establish which CCT methodology is most efficacious as well as how to deploy it in a way to elicit continued engagement (eg, increasing motivation or incentives and improved gamification). Moreover, more work is needed to identify the optimal method for combined delivery of CCT with other interventions, such as rTMS, particularly given that some methods (eg, required dose and most beneficial timing) may differ from CCT alone.

The studies presented here have some limitations and reveal several opportunities for future work. First, given that these were pilot studies focused on the safety, feasibility, and acceptability of combined accelerated iTBS and CCT, the intervention components were not explicitly dosed for efficacy. However, as higher doses (ie, increased treatment demands or durations) may impact participants’ engagement, continued consideration of feasibility and acceptability in future efficacy trials is necessary. Relatedly, the design of these studies did not include sham conditions, randomization, or blinding of participants or experimenters for either accelerated iTBS or CCT. Future randomized controlled trials that assess additive or synergistic effects of rTMS (accelerated iTBS or otherwise) and CCT are therefore needed. Second, both groups were predominantly non-Hispanic White with high levels of educational attainment. To ensure that research is representative of the general population, it is imperative that future research include participants from diverse sociodemographic backgrounds with varying levels of technology familiarity (which was not explicitly assessed in these studies). Third, people with aMCI and a history of stroke often experience neuropsychiatric symptoms, most commonly depression [43,44]; however, participants in our studies did not have clinically significant neuropsychiatric symptoms. Given that depressive symptoms are associated with poorer adherence to multidomain behavioral interventions (including cognitive training) in older adults [45], feasibility and acceptability of combined rTMS and CCT may be lower in people with MCI and comorbid neuropsychiatric symptoms. Alternatively, as rTMS to the l-dlPFC is US Food and Drug Administration–approved to treat depression, and both rTMS [46] and CCT [47] may improve depression or apathy in MCI, engagement in combined rTMS and CCT may instead be high in those with MCI and depression if they experience such benefits from the treatment. Therefore, our findings warrant replication in people with MCI and comorbid neuropsychiatric conditions.

In conclusion, this study provides initial evidence that accelerated iTBS to the l-dlPFC interleaved with CCT during treatment is feasible and acceptable to many patients with MCI. However, more limited engagement in the follow-up phase suggests that the feasibility of self-directed CCT may vary across people with MCI and by CCT delivery format (eg, personal trainer vs. à la carte), and indicates that regular reminders may be beneficial. These findings suggest that personalized treatment approaches and methods to reinforce CCT may enhance feasibility and adherence. Our results can directly inform future trials aimed at optimizing treatments that combine neuromodulatory and CCT interventions. Subsequent randomized controlled trials that evaluate whether combined rTMS and CCT yields a synergistic or additive impact on MCI symptoms are warranted.