Research Of Intracerebral Hemorrhage Prediction Model Based On CT Angiography

Intracerebral Hemorrhage(ICH),also known as hemorrhagic stroke,is an acute cerebrovascular disease with high onset,morbidity,disability,and mortality.ICH severely threatens human health and affects people’s lives.Quality,at the same time,brings huge economic burdens and heavy burdens on the lives of patients,their families and society.At present,ICH has become a common concern of China and the world.The onset of intracerebral hemorrhage is rapid and the condition changes rapidly.More than 70% of patients can develop hematoma enlargement or involving the ventricles at an early stage.ICH mortality can reach 40% within one month of onset,and as high as 54% in one year.At the end of 6 months,80% of people still have disabilities.In view of the rapid onset of intracerebral hemorrhage and extremely high disability mortality,it is particularly important to predict and prevent intracerebral hemorrhage in advance.Therefore,determining the risk factors of intracerebral hemorrhage and establishing a prediction model of intracerebral hemorrhage is of great significance for the prevention of intracerebral hemorrhage in advance.In this paper,the CT image data of intracerebral hemorrhage is provided by neurosurgery of Southwest Hospital of Chongqing.We extract the corresponding eigenvalues as research data by the software,and use statistical hypothesis testing to verify the characteristics of the factors affecting intracerebral hemorrhage,and use machine learning to establish an intracerebral hemorrhage prediction model.The predictive model play a guiding role in the prevention and treatment of intracerebral hemorrhage.At the same time,our work lays the foundation for further study.The main work of this paper is as follows:1.Use the potency test to estimate the sample size.Due to resource and moral reasons,it is impossible to obtain an infinite sample size for the optimization of the key parameters of the model,but it is also necessary to ensure that the experimental sample has statistical significance.In this paper,the sample size is estimated using the effectiveness estimation.By deciding the sample’s performance value to derive a statistically significant number of samples,we are instructed to extract a sufficient number of samples.2.A statistical test method was used to test the significance of the factors that may affect intracerebral hemorrhage.In order to determine which factors have influence on intracerebral hemorrhage,we performed statistical tests on the intracerebral hemorrhage data and non-intracerebral hemorrhage data by using T test and rank sum test to determine the characteristics of differences in mean values,and use the data as the research data for further research.3.The hybrid ensemble learning method(HELM)is used to establish a prediction model of intracerebral hemorrhage.The HELM algorithm is based on the traditional ensemble learning method Adaboost.This algorithm combines the traditional homogeneous ensemble learning method and the heterogeneous ensemble learning method.First,different types of base classifiers are integrated by using Adaboost algorithm,and then get the resulting classifiers for heterogeneous integration,and finally get our hybrid ensemble learning method.We use the hybrid ensemble learning method to train some of the extracted data to obtain a prediction model of intracerebral hemorrhage,and then use the remaining data to test the intracerebral hemorrhage prediction model and compare it with other methods to ensure the effectiveness of the intracerebral hemorrhage prediction model.In summary,this paper uses statistical testing methods to determine the characteristics of factors that may affect intracerebral hemorrhage,and uses hybrid ensemble learning methods to train and test on these characteristic data to establish prediction models for patients with intracerebral hemorrhage.By comparing with related algorithms and models,we prove the validity and stability of the intracerebral hemorrhage prediction model.