A multicenter, prospective observational study (INDEX study) using the ReoGo-J robotic system to improve paralyzed upper extremity (UE) function after stroke, involving 300 participants across 26 Japanese hospitals/clinics. Data sharing agreements are required for anonymized data access within two years of publication. The study protocol includes adjusting training tasks, reach ranges, and assistance modes based on UE paralysis severity, with compensatory movement monitoring to optimize rehabilitation. Statistical analysis employs item response theory to assess patient ability and UE function, with sub-analyses confirming model assumptions and local independence.