This study aimed to establish an artificial intelligence algorithm model for patients with Ischemic stroke(IS) based on general patient characteristics, drug treatment methods, and recovery effects. We sought to achieve individualized precise thrombolytic therapy and provide data support for clinical drug decision-making. Data of 55621 patients diagnosed with IS were extracted from the big data server of Yidu Cloud. After screening using the inclusion and exclusion criteria, 1236 and 619 patients were included in the neurological improvement and control groups, respectively for a total of 1855. Single-factor screening was conducted for different influencing factors within the scope of the investigation. Principal component analysis was used to reduce the influencing factor dimensions. We built logistic, support vector machine, C5.0 decision tree arithmetic, deep neural network, and wide and deep models. We further evaluated these models to compare their prediction performance, determine the best model algorithm, and identify the best model parameters. Feature engineering was used to construct a simplified model and evaluate its accuracy. Patient data extracted from another hospital in Dalian were used for external model validation. We included 1855 patients and 26 patient characteristics were included in the model through single-factor screening. The dimensions were reduced to two principal components principal component analysis, and the cumulative variance contribution rate was 93.1%. The wide and deep models showed the best prediction performance with an accuracy and F-index of 81.5% and 87.1%, respectively. Furthermore, the area under the receiver operating characteristic curve values of the training and test set were 0.753 and 0.793, respectively. The number of hidden layers and neurons in each layer of the model was seven and 15, respectively. Using Sigmoid as the activation function showed that the model parameters were optimal. Feature-engineering treatment showed that the accuracy of the model was 81.9%, and that of the external verification model after model simplification was 80.1%. The external verification accuracy of the wide and deep models was 75%, indicating good prediction performance and generalizability. The evaluation indexes of the wide and deep models were excellent. The artificial intelligence algorithm enabled the timely and effective provision of support for drug thrombolytic treatment planning for clinicians and references for clinical decision-making. Furthermore, the algorithm could potentially improve patient treatment outcomes, achieve personalized and accurate treatment for IS patients, and positively contribute to significantly reducing the social burden of disease.