To investigate the value of the predict recovery potential (PREP) 2 algorithm for prognostic stratification in the acute phase of stroke and to longitudinally analyze functional recovery trajectories across different prognostic subgroups in a Chinese stroke population.

In this prospective longitudinal study, 65 stroke patients admitted to the First Affiliated Hospital of Wenzhou Medical University within 3 days of onset between March 2023 and January 2024 were enrolled. The Shoulder Abduction and Finger Extension (SAFE) score was assessed within 3 days post-stroke, and transcranial magnetic stimulation-motor evoked potential (TMS-MEP) assessment was performed within 10 days (T1) for prognostic stratification. Prognostic stratification was subsequently conducted using the PREP2 algorithm. Functional assessments, including the upper limb Fugl-Meyer assessment (UE-FMA), national institutes of health stroke scale (NIHSS), and modified Barthel index (MBI), were conducted at T1, 1 month (T2), and 3 months (T3) post-stroke. All patients received standardized comprehensive rehabilitation therapy.

The PREP2-predicted stratification showed agreement with the 3-month outcome levels defined by UE-FMA recovery rate (weighted Kappa=0.785, P<0.001). Repeated-measures ANOVA revealed significant main effects of time and significant time-by-stratification interaction effects for both UE-FMA and MBI scores (P<0.001), indicating distinct functional recovery trajectories among the prognostic subgroups. Multiple linear regression identified the SAFE score (β=0.407, P<0.001) and MEP status (β=0.304, P<0.001) as independent positive predictors of UE-FMA scores at T3, whereas age (β=-0.181, P<0.001) and NIHSS score at T1 (β=-0.205, P<0.001) were negative predictors.

Under real-world clinical conditions in China, the application of the PREP2 algorithm in the early post-stroke phase (within 10 days) enables effective stratification to distinguish patient subgroups with different recovery trends.