Brain stroke continues to be a major contributor to global mortality and disability, necessitating advanced techniques for early diagnosis, prognosis, and personalized treatment strategies. This comprehensive review explores the advancements in interpretable machine learning (ML) models for stroke biomarker discovery. By improving model interpretability using techniques like Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP), decision trees, and attention mechanisms in neural networks, these models become more actionable in clinical practice. Applications in early stroke diagnosis, prognostic biomarker identification, stroke subtype stratification, treatment response prediction, and neuroimaging biomarker analysis are discussed. The review also highlights the importance of suitable ML techniques and existing datasets crucial for stroke research, including ISLES, ADNI, and SITS-ISTR. Despite challenges such as data limitations and the trade-off between accuracy and interpretability, ongoing advancements promise to bridge the gap between predictive accuracy and clinical usability. The advancement of personalized medicine and the improvement of patient outcomes can be significantly enhanced by explainable machine learning models, ultimately enhancing stroke care by addressing these challenges.