Development of a Structured Framework for Personalized Cognitive-Motor Exergame Training in Frailty

Abstract

Introduction: Aging is associated with progressive declines in physical and cognitive function, increasing the risk of frailty and loss of independence. Exergaming can offer combined cognitive-motor training, showing promise to counteract those declines. However, systematic frameworks for personalized and progressive exergame training are lacking. This study aimed to develop a structured, domain-based cognitive-motor exergame training framework for frail older adults and to specify how it is personalized and progressed.

Methods: The training plan was built on an exergame platform offering twenty-two cognitive-motor games. Each game was characterized by its predominant cognitive and/or physical properties and by the specific movement patterns required to play it. Based on these properties, games were grouped into three main domains (Cognition, Balance, and Endurance) reflecting multicomponent exercise recommendations. Thirteen multidirectional stepping games aligned most closely with neurophysiological test paradigms and formed the “Cognition” domain; two stepping-in-place games formed “Endurance”; six games emphasizing postural control formed “Balance”. Domains were further divided into subdomains to capture task-specific characteristics and allow gradual progression. Cognition included Memory, Reaction, Inhibition, Task Switching, and Visuospatial Orientation; Balance included Mediolateral Weight Shifting, Multidimensional Weight Shifting, Free Walking/Stepping, and Fine-Tuned Motor Coordination; Endurance had no subdomains. Where available, adjustable game parameters were used to create multiple “versions” supporting graded difficulty. The session structure followed established training principles and frailty guidelines: a brief aerobic warm-up, alternating stepping and weight-shifting/walking blocks to vary muscle activation and prevent overexertion, and a cool-down emphasizing relaxation and self-paced movement. Personalization translated physical and cognitive assessments into starting levels per subdomain, ensuring individualized training loads with age- and sex-specific adjustments where appropriate. For example, “Endurance” subdomain was based on percentage of expected 6-Minute Walk Test distance; “Reaction” subdomain on age-stratified Deary-Liewald Choice Reaction Time (RT); “Inhibition” subdomain on Go/No-Go RT; and “Balance” subdomain on Berg Balance Scale, Short Physical Performance Battery, Four Square Step Test and Time Up-and-Go dual-task combinations. Most subdomains had three levels (1–3), while highly complex or minimally adjustable tasks began at the easiest level to ensure feasibility. Personalization was continuously refined through regular trainer–participant calls, allowing dynamic adjustment of game choices and settings. Progression was implemented at two layers. In-game progression (built-in) used each game’s dynamic difficulty features (e.g., faster stimuli, denser obstacles) to scale challenge within a session. Inter-game progression ordered activities from easiest to hardest, using principles such as increasing duration, step-direction complexity (lateral → anterior → posterior), stimulus density, or built-in difficulty percentages. Advancement was governed by a 10-point visual analog scale rule after each activity: progress up if VAS≤3 (too easy), step down if VAS≥7 (too hard), otherwise repeat; trainers could override based on safety and clarity.

Discussion: This structured framework provides a comprehensive, evidence-based approach to personalized and progressive exergame training for frail individuals. We are now in the process of assessing its validity through longitudinal game performance data and objective verification of domain classifications. Establishing such data-driven personalization methods could inform AI-based adaptive training systems that optimize both cognitive and motor outcomes in aging populations.