Intelligent Diagnosis and Treatment Model of Hemorrhagic Stroke Based on Graph Convolutional Neural Network

In the prevention and treatment of hemorrhagic stroke, it is of great significance to judge and select influencing factors as well as make treatment plans through modern intelligent medical methods based on machine learning algorithm. In this paper, we mainly focus on the development, treatment and prognosis of hemorrhagic stroke. Firstly, for the risk of hematoma expansion, an algorithm is used based on the recorded data, and features selection is done to predict whether patients would develop hematoma expansion within 48 hours after the disease. Secondly, in view of the occurrence and progression of edema around hematoma, time-dependent modeling and cluster-based iterative methods are carried out, and subgroups are divided according to the edema volume progression pattern. Last but not least, the prognosis of the patients (90-day mRS) after disease is predicted based on the Graph Convolutional Neural Network (GCN) designed by us. Our proposal won the second prize in the 2023 "Huawei Cup" Chinese Postgraduate Mathematical Contest in Modeling (ranked in the top 2%).