Abstract / Summary
Virtual reality may improve cognition and neuropsychiatric symptoms in mild cognitive impairment and dementia, but the comparative effects of different intervention strategies remain unclear. We conducted a systematic review and network meta-analysis of randomized controlled trials identified from nine databases from inception to August 6, 2025. Thirty-six trials involving 2,048 adults with mild cognitive impairment or dementia were included. Interventions lasted 4–24 weeks, with sessions of 18–100 min. Challenge-based semi-immersive virtual reality ranked highest for global cognitive function (SUCRA 80.9%), followed by problem-solving fully immersive virtual reality (SUCRA 70.6%). Challenge-based fully immersive virtual reality ranked highest for executive function (SUCRA 75.7%) and memory (SUCRA 76.4%). Network meta-regression identified no significant moderation by participant or intervention characteristics. Virtual reality, particularly challenge-based semi-immersive and fully immersive approaches, may offer cognitive benefits, although methodological heterogeneity and risk of bias limit certainty.