A Computerized Intelligent Diagnostic Model for Hemorrhagic Stroke Based on LSTM Neural Network Prediction

In this paper, we conducted an in-depth study on the diagnosis and prognosis prediction of hematoma dilatation and peripheral edema after hemorrhagic stroke. First, using the data extracted from the table such as the flow number of the imaging examination, a discriminative model of hematoma expansion was established, which discriminated against hematoma expansion based on explicit conditions and successfully realized the prediction of the probability of hematoma expansion with an accuracy rate of nearly 74%. Secondly, this paper analyzed the relationship between edema volume and imaging time, used K-means clustering to classify the edema volume, and explored the effect of different treatment methods on the progression of edema volume. Finally, this paper used LSTM neural network to successfully construct a prediction model for MRS score, which further optimized the accuracy of diagnosis and treatment.