Efficacy of digital interventions for patients with post-stroke aphasia: a systematic review and network meta-analysis

Abstract

Background Digital interventions show potential for improving language function in post-stroke aphasia (PSA), but contradictory evidence leaves the optimal strategy unclear. Objective This study aims to evaluate the effects of different digital interventions on language function and quality of life in patients with PSA, and to provide evidence-based support for clinical rehabilitation decisions. Methods A systematic search across PubMed, Web of Science, Cochrane Library, Embase, Scopus, and China National Knowledge Infrastructure (CNKI) (inception to May 2026) identified randomized controlled trials (RCTs) evaluating digital interventions (e.g., virtual reality [VR], computer-assisted language therapy [CALT], telerehabilitation [TR]) for adults with PSA. Eligible studies reported Western Aphasia Battery or Stroke and Aphasia Quality of Life Scale-39 (SAQOL-39) outcomes. Data were extracted independently. Risk of bias was assessed using the Cochrane Risk of Bias 1 (RoB 1) tool. Network meta-analysis was performed using R 4.5.2, incorporating a consistency model, a random-effects model, and I2statistics to assess heterogeneity. Sources of heterogeneity were analyzed through sensitivity analyses, subgroup analyses, and meta-regression. Intervention efficacy was ranked using Surface Under the Cumulative Ranking Curve (SUCRA). Robustness was validated via the Egger test and sensitivity analyses. Evidence quality was assessed using the GRADE framework. Results Twenty-five RCTs involving 1,069 participants were included. Compared to conventional therapy (C), VR showed significant superiority in aphasia quotient (mean difference [MD] = 10.77, 95% CI 9.48 to 12.06; SUCRA = 89.57%), spontaneous speech (MD = 2.35, 95% CI 1.71 to 2.99), auditory comprehension (MD = 1.47, 95% CI 0.81 to 2.13), and repetition (MD = 1.98, 95% CI 1.19 to 2.76); Ranking was VR > CALT>TR > C. For naming, CALT ranked highest (MD = 1.41, 95% CI 0.79 to 2.03; SUCRA = 89.87%); Ranking was CALT>VR > C > TR. For SAQOL-39, VR showed the best effect (MD = 0.51, 95% CI 0.33 to 0.70; SUCRA = 96.40%). Ranking was VR > TR > C > CALT. Conclusion Digital interventions effectively improve language function in PSA. VR demonstrated potential advantages for aphasia quotient, speech, comprehension, repetition, and SAQOL-39, whereas CALT provides targeted advantages for naming impairments. Caution is warranted due to study heterogeneity, bias risks, and limited sample sizes. Future research should optimize intervention protocols, integrate neuromodulation or traditional therapeutic techniques, and explore cost-effective implementation strategies. Systematic review registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420261373199, Identifier CRD420261373199.