New coronal pneumonia patient outcome prediction method based on interpretable machine learning algorithm

Abstract

The invention provides a new coronal pneumonia patient outcome prediction method based on an interpretable machine learning algorithm, wherein the method comprises the steps: extracting COVID-19 patient data from a database, and dividing the patient data into an experimental group and a control group according to the illness state conversion condition of a patient; interpolating the missing value of each index through random forest regression; screening the indexes of the input model, and taking the screened indexes as key risk factors for identifying the deterioration of the patient; inputting the key risk factors of the patient into the XGBoost model and the logistic regression model; selecting an XGBoost model with better prediction expressive force to generate an index combination, and performing prediction by using the XGBoost model and recording the prediction result; defining the early warning range of the key index; when the key risk index of the patient enters the early warning range, giving out an alarm prompt to medical staff. According to the invention, the calculation result of the algorithm and the clinical experience of a doctor are synthesized, and two index combinations composed of 15 first groups of indexes and 5 second groups of indexes are proposed to be used for predicting the condition of the new coronal pneumonia patient.

Description

technical field [0001] The present invention relates to the field of machine learning technology, in particular to a method for predicting the outcome of patients with new coronary pneumonia based on interpretable machine learning algorithms. Background technique [0002] The surge in the number of cases of novel coronavirus disease (COVID-19) infection is a huge challenge to the management of medical resources. Although approximately 81% of COVID-19 patients exhibit mild or moderate symptoms, some patients experience sudden deterioration, rapidly becoming severe or critically ill. Therefore, early intervention for the deterioration of COVID-19 patients will greatly help the management of patients and the allocation of medical resources. In the process of realizing the present invention, the inventor finds that there are at least the following problems in the prior art: [0003] 1) Among the published studies on the poor prognosis of COVID-19, most of them still use statis...

AI Technical Summary

Problems solved by technology

No researchers currently focus on state transitions for disease progression

[0005] 3) Although machine learning has achieved good prediction results in existing research, the number of indicators required by the model is large and the sampling is complex, including various laboratory indicators, and it takes a long time to obtain all the indicators required by the model. index

Ignores the issue of indicator availability and timeliness when using machine learning models in real-world contexts

[0006] 4) At the same time, no matter whether it is based on traditional statistical methods or machine learning methods, it only identifies the risk factors that lead to the deterioration of the patient's condition, hospitalization or final death, but does not provide the corresponding warning range of the indicators

Images