The use of machine learning (ML) in predicting disease prognosis has increased, and researchers have adopted different methods for variable selection to optimize early screening for AIS to determine its prognosis as soon as possible. We aimed to improve the understanding of the predictors of poor functional outcome at three months after discharge in AIS patients treated with intravenous thrombolysis and to construct a highly effective prognostic model to improve prediction accuracy. And four ML methods (random forest, support vector machine, naive Bayesian, and logistic regression) were used to screen and recombine the features for construction of an ML prognostic model. A total of 352 patients that had experienced AIS and had been treated with intravenous thrombolysis were recruited. The variables included in the model were NIHSS on admission, age, white blood cell count, percentage of neutrophils and triglyceride after thrombolysis, tirofiban, early neurological deterioration, early neurological improvement, and BP at each time point or period. The model's area under the curve for predicting 30-day modified Rankin scale was 0.790 with random forest, 0.542 with support vector machine, 0.411 with naive Bayesian, and 0.661 with logistic regression. The random forest model was shown to accurately evaluate the prognosis of AIS patients treated with intravenous thrombolysis, and therefore they may be helpful for accurate and personalized secondary prevention. The model offers improved prediction accuracy that may reduce rates of misdiagnosis and missed diagnosis in patients with AIS.