Ying Liu, Qian Hu, Shao Yin, Yue Zhong, Qiang Ji, Fengya Zhu
Robot-assisted rehabilitation may improve selected balance-related outcomes after stroke, particularly BBS. However, current evidence is limited by low methodological quality, unclear or high risk of bias, low or very low certainty, heterogeneity, and primary-study overlap. Robot-assisted rehabilitation should be considered a potential adjunct to comprehensive post-stroke rehabilitation, and higher-quality evidence is needed to clarify its effects on balance and postural control.
OBJECTIVE: Balance and postural-control impairments are common after stroke and can restrict mobility and independence, but the effectiveness of robot-assisted rehabilitation for these outcomes remains uncertain. This umbrella review aimed to summarize and critically appraise evidence from systematic reviews on robot-assisted rehabilitation for balance function and postural control after stroke.
METHODS: This umbrella review was prospectively registered in PROSPERO (CRD420261426090). PubMed, Embase, the Cochrane Library, and Web of Science were searched from inception to March 30, 2026. Eligible publications were systematic reviews evaluating robot-assisted rehabilitation in adults after stroke and reporting at least one balance- or postural-control-related outcome. Methodological quality, reporting completeness, risk of bias, certainty of evidence, and primary-study overlap were assessed using AMSTAR-2, PRISMA 2020, ROBIS, GRADE, and corrected covered area (CCA), respectively.
RESULTS: Twelve systematic reviews were included. The included reviews evaluated robot-assisted gait training, Lokomat, overground robot-assisted gait training, end-effector robots, exoskeleton robots, lower-limb rehabilitation robots, and mixed robotic devices. The Berg Balance Scale (BBS) was the most frequently reported outcome. Several reviews suggested favorable effects of robot-assisted rehabilitation on BBS, whereas findings for Timed Up and Go, Fugl-Meyer balance, Rivermead Mobility Index, and instrumented postural-control outcomes were less consistent. AMSTAR-2 rated 10 reviews as low quality and two as critically low quality. ROBIS indicated one review with high overall risk of bias and the remaining reviews with unclear risk of bias. The certainty of evidence for major outcomes was generally low or very low. Primary-study overlap was high for overall balance-related evidence (CCA = 12.06%), BBS evidence (CCA = 11.61%), and TUG/TUGT evidence (CCA = 11.76%).
CONCLUSION: Robot-assisted rehabilitation may improve selected balance-related outcomes after stroke, particularly BBS. However, current evidence is limited by low methodological quality, unclear or high risk of bias, low or very low certainty, heterogeneity, and primary-study overlap. Robot-assisted rehabilitation should be considered a potential adjunct to comprehensive post-stroke rehabilitation, and higher-quality evidence is needed to clarify its effects on balance and postural control.
SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261426090, PROSPERO (CRD420261426090).