Cognitive impairment (CI) occurs in 30% to 70% of people with multiple sclerosis (MS) and has a profound influence on a patient’s personal functioning, social interaction, employment, and overall quality of life [1,2]. Being able to effectively detect CI is essential to better managing further decline [3] and helping patients navigate problems related to their daily living. The most commonly affected cognitive domains are processing speed, attention, executive function, and memory [4,5]. Currently, clinical cognitive assessment relies on a comprehensive neuropsychological examination, which is time-consuming and extensive. The examination results may be affected by patient fatigue or loss of engagement. Given the large interindividual variability in the pattern of CI in MS [6,7], the traditional nonadaptive assessments may overlook cognitive deficiencies [8]. Taking advantage of technology, digital tools offer a platform to integrate personalizing features, including adaptive staircase algorithms to video game–style mechanics for cognitive assessment [8-15]. This approach can mitigate potential ceiling and floor effects when interindividual variability is high, and leads to more reliable assessments that can be completed in a timely manner [8,16]. Furthermore, digital tools can be easily applied in different settings, including patients’ homes, which substantially improves health care accessibility for patients who have difficulties with travel to the clinic due to cognitive or physical disabilities [14,17,18].

Digital tools have been used for functional assessment, rehabilitation, and health care monitoring in clinical populations such as stroke [19,20], schizophrenia [21], depression [10,12], attention-deficit/hyperactivity disorder [15,22-24], and neurodevelopmental disorders [8,9]. In MS, studies have demonstrated that by using digital tools, health care services can be delivered effectively [25,26] and even that comprehensive neurological exams can be performed remotely [27]. However, to the best of our knowledge, applications of serious video game–based digital tools incorporating closed-loop adaptation mechanics as cognitive assessments for people with MS have not been assessed.

We previously evaluated whether a video game–based digital tool, EVO-AKL-T01 (Akili), could improve cognition in MS [17,18]. After these studies, we looked to better characterize cognitive function using similar tools based on these cognitive therapeutics. EVO Monitor (Akili) was developed based on findings that a precursor (NeuroRacer), embedded with adaptive algorithms, is sensitive to age-related cognitive decline across the lifespan and can enhance cognitive control [28]. EVO Monitor, which includes some of NeuroRacer’s early closed-loop features, is a novel, tablet-based, digital platform that is incorporated with video game mechanisms, visual and auditory feedback, adaptive algorithms, and sophisticated graphics that is designed to assess executive function, attention, and information processing speed for clinical populations. Specifically, it consists of 3 tasks: perceptual discrimination, visuomotor tracking, and multitasking. In the perceptual discrimination task, the participants complete a go/no go–like paradigm, in which they tap the iPad screen for correctly colored target stimuli while ignoring distracting targets. The visuomotor tracking task requires the participant to tilt the iPad to steer an avatar around obstacles. The multitasking task requires participants to perform both perceptual discrimination and visuomotor tracking concurrently. The tool is designed to challenge attention, goal management, and information processing speed in the setting of interference. EVO Monitor has been validated as a tool to assess cognitive ability in children with and without neurodevelopmental disabilities. For example, it can differentiate between the performance of children with or without 16p11.2 BP4-BP5 deletion and healthy controls, whereas traditional nonadaptive cognitive assessments overlook group differences [8].

In this study, we aimed to determine the feasibility and potential assessment sensitivity of EVO Monitor as an unsupervised, tablet game–based approach with adaptive algorithms to assess cognition in MS. In this context, 100 people with MS and 24 adults without MS were tested with EVO Monitor and standard measures, including the written version of Symbol Digit Modalities Test (SDMT), the most sensitive test for detecting cognitive involvement in the MS course [29,30]. Group-level differences were examined to evaluate whether EVO Monitor is sensitive to differences among participants with MS with and without CI as well as participants without MS. The association between performance in EVO Monitor and standard measures was investigated. Furthermore, since cognition in MS has been shown to correlate with physical fitness, where aerobic capacity and muscular strength outcomes are associated with cognitive processing speed and inhibitory control [31,32], and magnetic resonance imaging (MRI) measures have shown that thalamus damage is associated with the presence of CI, and frontal lesion is associated with executive function [33,34], exploratory analyses of the correlation between performance in EVO Monitor and both physical measures and MRI volumetric data (thalamus and frontal lobes) were conducted to understand whether cognitive performance assessed by a video game–based digital tool (EVO Monitor) would be associated with physical activity and structural MRI measures.

In this study, we aimed to determine whether in-game assessment features of EVO Monitor, an unsupervised, digital, video game–based tool integrated with adaptive algorithms, could represent a novel and sensitive way to perform unsupervised cognitive evaluations in MS. We found a significant group-level difference in performance on EVO Monitor among participants with MS with and without CI as well as participants without MS. Furthermore, we discovered an association between performance in EVO Monitor and standard cognitive measures, physical measures, and MRI volumetric data. These results provide evidence that EVO Monitor, an unsupervised, tablet game–based digital tool designed to assess executive function, attention, and information processing, could effectively assess cognitive performance in people with MS.

We demonstrated that EVO Monitor is sensitive enough to reveal group-level differences, as shown by a lower threshold level in participants with MS with CI compared to both participants with MS without CI and participants without MS in all 3 task conditions. The EVO Monitor program was developed to challenge nearly all aspects of cognition, including executive function, attention, and information processing. In perceptual discrimination and visuomotor tracking tasks, fast information processing is highly demanded, given that the participants are asked to make a fast response to certain stimuli (ie, colored target) and rapidly changing environment (ie, dynamically moving road with walls and obstacles), respectively. To successfully perform the multitasking challenge, there are additional cognitive requirements for selective attention, sustained attention, task switching, and goal management. Since processing speed, attention control, and executive function are the most commonly affected cognitive domains in MS [4,5], the lower EVO Monitor performance in participants with MS and CI may reflect a lower performance in these cognitive domains in general. These results suggest that cognitive dysfunction in MS can be captured by EVO Monitor, a digital tool with adaptive algorithms developed for cognitive assessment.

Using unsupervised digital tools provides a less stressful context and a high level of standardization of assessment. Moreover, it can be easily applied in multiple settings, including patients’ homes, which substantially improves access to these therapies for patients who may face scheduling, geographic, or economic barriers in accessing standard forms of cognitive assessment. With the advances in digital therapeutics, assessment and health care services for people with MS have been transitioning to digital platforms [25,26,52]. These results extended the application of digital tools for cognitive assessments in MS by incorporating advanced visualization and reward loops with closed-loop adaptation mechanics to enhance engagement and reduce interindividual variability, which leads to more reliable assessments [8,47].

An association between performance in EVO Monitor and standard measures was observed. Participants with a better performance in EVO Monitor showed higher SDMT scores. SDMT is considered the most sensitive measurement for the evaluation of cognitive involvement and information processing speed in the MS course [29,30]. While the participant substitutes geometric symbols for numbers while scanning a response key, a number of domains—processing speed, sustained attention, visual scanning and tracking, and motoric quickness—are being challenged. Although the task structures are different between EVO Monitor and SDMT, the shared cognitive functions subserving the two tasks may explain the correlational links. The results suggest that EVO Monitor can reliably reflect performance in cognitive tasks that involve information processing speed, selective attention, sustained attention, visual tracking, and goal management. Importantly, analyses including only participants with MS did not change the results but further revealed a negative correlation between clinical characteristics (ie, EDSS and disease duration) and performance in EVO Monitor, where participants with MS who had a longer disease duration and higher level of disability showed a lower performance in EVO Monitor. Although the correlation was modest, given that cognitive deficits tend to progress with disease duration [53] and disability progression [54] in MS, the observed association was expected.

In an exploratory analysis, we discovered an association between EVO Monitor performance and the T25FW as well as the bilateral rostral middle frontal and thalamic volume. In line with previous studies reporting that cognition in MS is associated with physical fitness and balance [55,56] and the concept of cognitive-motor coupling [57], in this study we found that better EVO Monitor performance is correlated with a faster walking speed as measured by the T25FW. Imaging [58] and postmortem [59] studies have suggested that frontal and subcortical regions involved in executive function and cognitive processing speed are also related to the spatial and temporal aspects of gait. The observed association between EVO Monitor performance and the T25FW may be explained by cerebral injury causing impairment in both domains in MS.

Additionally, the performance in the perceptual discrimination task of EVO Monitor positively correlated with rostral middle frontal and thalamic volume. The rostral middle frontal region is part of the dorsolateral prefrontal cortex, an area that is considered the center of executive function [60] and several domains of cognitive control abilities required to perform the EVO perceptual discrimination task, including sustained attention [61], selective attention [62], and inhibitory control [63]. Studies have also demonstrated that the thalamus mediates arousal states [64], and thalamic atrophy is the most significant MRI correlate of CI in MS [65,66]. It is unclear why performance in multitasking, a task including both perceptual discrimination and visuomotor tracking, did not correlate with thalamus and frontal volumetric data. Given that the posterior cortex has been linked to sustaining attention to spatial locations [67], a key component of the visuomotor task, it is possible that the multitasking task does not solely rely on frontal and thalamic resources but is also supported by the posterior cortex. Therefore, the examined correlation was not strong enough to be detected, as it was in the perceptual discrimination task.

Although our analyses were exploratory, the observed correlations support the need to better understand how EVO Monitor performance is associated with physical performance and the structural and functional changes of the brain in people with MS. Specifically, studies with functional MRI and connectivity data would provide essential information about neural correlates of the EVO Monitor tasks and, more importantly, pathological changes related to CI and neural plastic changes as a result of cognitive remediation.

Previously, we showed that EVO-AKL-T01, a video game–based digital tool similar to EVO Monitor, is an effective in-home cognitive remediation program for MS [17,18]. The high adherence rate during the 4- to 6-week home-based cognitive rehabilitation strategy indicated that remote digital tools are well accepted by patients with MS, who may have limited access to cognitive assessment or treatment. Since EVO Monitor is a digital tool designed as a self-guided assessment, it can be used as an in-home cognitive evaluation with multiple assessments to track either the progression of CI or the responsiveness to cognitive interventions, which can substantially help patients navigate problems related to cognitive issues. Future studies are warranted to evaluate the use of EVO Monitor in everyday situations across different contexts and investigate whether the results would be different due to fluctuations in cognitive ability during the day. Another key factor in cognitive assessment is patient fatigue. Well-designed studies that control for impacts of perceived fatigue and fatigability on cognitive assessments measured by digital tools are needed. Future studies comparing how different aspects of fatigue (eg, physical vs cognitive fatigue) affect cognitive performance as assessed by a comprehensive neuropsychological examination and by a digital tool that provides a shorter assessment time would provide additional insight into the role of fatigue in digital cognitive assessments.

There are some limitations to this study. Since we only included one time point in this cross-sectional analysis, it is difficult to determine the reproducibility of the observed results to conclude the test-retest reliability of EVO Monitor. In the exploratory analysis of the association between performance in EVO Monitor and MRI volumetric measures, clinically acquired MRI scans were only available for about half of the participants with MS (n=56), and the impact of acquisition protocol heterogeneity on our MRI metrics should be taken into account while a robust image processing pipeline was applied. The age difference between the non-MS group and non-CI MS group is one of the caveats in this study. However, in our analyses, age was included as a covariate to control for the potential influence of age difference on the results. Skills using digital tools may be a confounding factor influencing the results, as participants with better digital tool skills or more experience using tablet devices may have performed better. Future studies investigating digital tools should control for participants’ experience and skills using digital devices.

This study extended the application of digital tools for cognitive assessments in MS by incorporating built-in adaptive staircase algorithms to enhance engagement and mitigate interindividual variability. Furthermore, the encouraging findings suggest that EVO Monitor, an unsupervised, tablet game–based program, is a clinically valuable approach to capturing CI in MS. Since CI is one of the most debilitating manifestations of MS and traveling to clinics may be burdensome due to deficits in mobility or cognition, some patients may have limited access to cognitive assessments. The application of digital cognitive assessments provides flexibility, as the testing can be performed in different settings, including patients’ homes. The development of digital cognitive assessments helps patients effectively detect CI and navigate cognition-related problems in their daily living. Future studies with multiple points of data collection and a deeper investigation of how physical performance and the functional and structural changes of the brain affect cognitive performance as measured by digital tools are warranted to provide additional insight.