This paper presents an in-depth study on the prediction of clinical prognosis and feature correlation in patients with hemorrhagic stroke. Utilizing a publicly available dataset on hemorrhagic stroke, a comprehensive model was established to predict the clinical prognosis of patients with hemorrhagic stroke by analyzing clinical data and imaging information, and the importance of associated characteristics was analyzed. Experimental results demonstrate that the random forest algorithm performs well in predicting multi-class problems, accurately forecasting patient prognosis, and determining the importance of each feature in relation to the target variable (90-day mRs scores). Age, blood pressure, time interval between onset of illness and first imaging, and imaging-related features were identified as important influencing factors. These findings provide valuable insights for clinical diagnosis and treatment.