Critical patient prognosis prediction method

Abstract

The invention provides a critical patient prognosis prediction method, provides a critical patient prognosis prediction method based on an ensemble learning method, and belongs to the field of critical patient prognosis evaluation. The method comprises the steps that related data of a patient and a hospital are included through a structured language; feature variables are screened based on a random forest, and a prediction model is constructed through regularization logistic regression, K nearest neighbor, a support vector machine, the random forest, limit gradient lifting and a deep neural network algorithm; and through prediction probabilities of different methods, ensemble learning is carried out through a limit gradient lifting algorithm to finally predict whether the critical patient is dead or not, and the occurrence probability of the critical patient is further calculated. According to the method, patient data and hospital feature information are utilized as much as possible, the prognosis of the critical patient is predicted in an individualized mode, prediction bias caused by a certain model is weakened, and prediction accuracy is improved.

Description

technical field [0001] A method for predicting the prognosis of critically ill patients, especially a method for predicting the prognosis of critically ill patients based on integrated learning of multiple machine learning methods. Background technique [0002] Critically ill patients usually have life-threatening organ or system dysfunction, and early assessment and active treatment are crucial to saving the lives of patients. The condition of critically ill patients is complex and changes rapidly; therefore, it is difficult to assess the risk of death of patients based on subjective experience alone. At present, some scoring systems are used clinically to reflect the severity of critically ill patients and have certain value in predicting death, including simplified acute physiology scoring models, acute physiology and chronic health status scoring models, and death probability models. [0003] These models are all based on the logistic regression method and have some sho...

AI Technical Summary

Problems solved by technology

[0003] These models are based on the logistic regression method and have some shortcomings, including ① the model requires a linear relationship between the predictor variables and the corresponding results

Although fractional polynomials can be used to fit nonlinear relationships, this requires the modeler to identify the effective representation of the predictor variables, so that the model has the best predictive performance; ②The existence of sharing will lead to the regression coefficient in the LR model The estimation of is unstable, making the model coefficients no longer interpretable

③The logistic model is sensitive to multicollinear data, missing data, and unbalanced data, and it is difficult to achieve high-order interactions

④External verification found that the performance of the existing model is not ideal

For example, mortality rates are overestimated when these models are used on data other than development

Images