Statistical Modeling of Prognosis of Cerebral Infarction Thrombolysis and Analysis of Influencing Factors
This paper investigates the influencing factors affecting the prognosis of throm-bolytic therapy in patients with cerebral infarction and combine the data to establish a pre-diction model to provide theoretical basis and support the development of future clinical treatment.Collecting cases of cerebral infarction patients from the First Hospital of Harbin Medical University and screening data indicators to establish a univariate analysis model,and the measurement data and count data were tested separately to obtain preliminary influenc-ing factors and to include the indicators with in the establishment of the machine learning prediction model.The results showed significant differences between the poor prognosis group and the good prognosis group in age(t=-3.050,P=0.003),post-thrombolytic glucose value(z=3.490,P《0.01),and previous hypertension(x2=6.853,P=0.009).The accuracy of the four machine learning prediction models constructed based on logistic regression,decision tree,random forest,and plain Bayesian algorithms were 79％,79％,75％,and 73％,respectively;the area under the curve was 0.85,0.81,0.78,and 0.81,respectively.The four constructed prediction models all had good prediction ability of whether the prognosis of patients with cerebral infarction was good or not,and had potential value for clinical research.
machine learning predictioncerebral infarction diseaseinfluencing factorsclin-ical research
NETL
NSTL
万方数据
