Cognitive frailty is a clinical syndrome in which cognitive impairment (e.g., poor memory, visuospatial function) and physical frailty (e.g., slowness, poor muscle strength, physical inactivity) co-exist. It is prevalent in community-dwelling older people. The progressive decline of cognitive and physical functions restricts older people from participating in activities (e.g., social get-togethers). Reduced participation further jeopardizes their life-space mobility (e.g., ability to travel to areas far away from home). Therefore, those with cognitive frailty are at risk of developing dementia and becoming dependent.

Simultaneous motor-cognitive training is more effective at promoting optimal functioning in older people than motor or cognitive training alone. Gaming is effective at promoting the motivation to participate. The contents of games in the market are unrelated to the context or daily living of the elderly. Currently, available training is non-simultaneous. This makes the training less transferable to the daily life of the elderly and reduces its effects.

Virtual reality (VR) technology can provide a virtual space that mimics the real environment. This allows clients to participate in daily activities in a virtual space. Older people can be trained to improve their cognitive and physical skills in a painless, fun way. However, the effect and feasibility of employing simultaneous motor-cognitive training launching on a VR platform mimicking the daily living environment in older people with cognitive frailty is poorly known.

Following the findings from the previous proof-of-concept test (registration number: NCT04467216), we proceed to implement the study to 400 participants from six different elderly centres between the period of March 2021 and December 2022.

In the intervention, VR will be employed to simulate a daily living environment familiar to older people. Participants will wear a commercially available head-mounted VR system with hand-held controllers to experience the participation of daily activities in a virtual environment. Simultaneous physical and cognitive training will be embedded in the training system to promote optimal function. Participants will attend physical training in a sitting position through cycling on an ergometer and moving the hand-held controllers. Motion sensors built into the VR system and ergometer will track these movements to control everyday tasks in the virtual environment (e.g., moving around the city). Simultaneously, participants will undergo cognitive training by participating in various tasks demanding cognitive functions, such as visual-spatial (e.g., wayfinding) and problem-solving (e.g., wallet loss) functions. Gamification will be employed to promote the motivation to participate. All training activities will be gamified by blending in fun elements, such as difficulty-levelling, competition, and e-tokens. Co-participation is allowed to promote interpersonal interactions. The prototype enables real-time co-viewing among participants. Other elderly centre members and the activity facilitators can share the view of the participants in the game on either a large-screen monitor or a tablet computer, allowing them to simultaneously discuss and share about their gaming experiences.

Inclusion Criteria:

• Age ≥ 60 years,
• Self-reported or informant-reported cognitive complaints
• Objective cognitive impairment, as defined by a Clinical Dementia Rating of 0.5 and a Montreal Cognitive Assessment (MoCA) score of <25
• Preservation of one's independence, as defined by the Lawton's Instrumental Activity of Daily Living score of >14
• No diagnosed dementia, as observed in the medical record
• Physical frailty from being pre-frail to frail, as defined by a Fried Frailty Index (FFI) score of 1-5.

Exclusion Criteria:

• Participants who have impaired mobility, as defined by Modified Functional Ambulatory Classification (MFAC) < Category 7 (i.e., Outdoor walker),
• or probable dementia, i.e., MoCA < 17 or clinical dementia rating ≥ 1.

Outcome assessors at both pre- and post-observations will be blinded to the group label.