We analyzed clinical and proteomics data from stroke patients aged ≥18 years enrolled in a prospective acute brain injury study (2014–2023, Yale University).  Our outcome was differential protein expression among patients with acute ischemic stroke (AIS), intracerebral hemorrhage (ICH), transient ischemic attack (TIA), and stroke mimics (MIM). Samples were collected within 24 hours of stroke onset before treatment. Plasma was depleted using High-Select Top 14 Protein Depletion Resin. We performed data-independent acquisition-based mass spectrometry (DIA-MS) proteomics on an Orbitrap Exploris 480. DIA-MS data were quantified using DIA-NN, log₂ transformed, batch corrected using Combat function (sva package in R), imputed for ≤30% missing values using Perseus, and normalized by total area sum. To identify top group discriminators, we performed feature selection by sparse partial least squares discriminant analysis (sPLS-DA) using mixOmics R package. Regularized LASSO regression identified classifiers for each stroke subtype, and model performance was assessed by 10-fold cross-validation using receiver operating characteristic curves. Pathway analysis was conducted using clusterProfiler package.