JMIR Publications

ABSTRACT

Background:

Early assessment of mild cognitive impairment (MCI) in older adults is crucial, as it enables timely interventions and decision-making. In recent years, researchers have been exploring the potential of gamified interactive systems (GIS) to assess pathological cognitive decline. Yet, researchers are still investigating effective methods for system integration, designing GIS that are perceived as engaging whilst also improving the accuracy in assessing cognitive decline.

Objective:

This review aims to comprehensively investigate GIS used to assess MCI. Specifically, we reviewed the existing systems to understand the different game types (including genres and interaction paradigms) employed for assessment. Additionally, we examined the cognitive functions targeted. Finally, we investigated the evidence for the performance of assessing MCI through GIS by looking at the quality of validation for these systems in assessing MCI and the diagnostic performance reported.

Methods:

A systematic search was conducted in IEEE Xplore, ACM Digital Library, and Scopus to identify interactive gamified systems developed for assessing MCI. Game types were categorized according to genres and interaction paradigms. The cognitive functions targeted by the systems were compared with those assessed in the MoCA. Lastly, we examined the quality of validation on ground truth, relevance of controls, and sample size. The diagnostic performance on sensitivity, specificity, and AUC are reported.

Results:

A total of 81 papers covering 49 GIS were included in this review. The primary game types used for MCI assessment were classified as casual games (n = 30), simulation games (n = 17), full-body movement games (n=4), and dedicated interactive games (n = 3). Only six out of 49 systems assess cognitive functions comprehensively, compared to those functions assessed via the MoCA. The reported diagnostic performances of GIS were comparable to common screening instruments like MMSE and MoCA, with some systems reporting near-perfect performance (100% sensitivity and specificity).

Conclusions:

This review provides a comprehensive summary of the literature on GIS for assessing MCI, explores the cognitive functions assessed by these systems, and evaluates their diagnostic performance. The results indicate that current GIS hold significant promise for the assessment of MCI, with several systems demonstrating diagnostic performance comparable to established screening tools. However, these systems' model training and validation exhibited significant deficiencies. Hence, despite some systems reporting impressive performance, there is a need for improvement in the validation studies, particularly concerning sample size and methodological rigor. Finally, we advocate for increased longitudinal research to enhance the reliability of these systems in evaluating MCI.