ICU (Intensive Care Unit) hospitalization duration and death risk prediction method based on convolutional neural network

Abstract

The invention discloses an ICU (Intensive Care Unit) hospitalization duration and death risk prediction method based on a convolutional neural network. The method comprises the following steps: acquiring basic data, processing and classifying to obtain a training data set, a verification data set and a test data set; constructing a basic prediction model based on the time sequence cavity divisible convolution with different receptive fields and a context perception feature fusion module; setting a loss function and training, verifying and testing the basic prediction model by adopting the data set to obtain an optimal prediction model; and performing ICU hospitalization duration and death risk prediction of actual personnel by adopting the optimal prediction model. According to the method, each feature is independently coded by using a time sequence hole divisible convolutional network, and a context sensing feature fusion method is provided; a final hospitalized person representation is generated by combining a multi-view and multi-scale feature fusion module for prediction; therefore, the method is high in reliability, high in precision and good in effect.

Description

technical field [0001] The invention belongs to the field of data processing, and in particular relates to a method for predicting the length of stay in an ICU and the risk of death based on a convolutional neural network. Background technique [0002] With the development of economy and technology and the improvement of people's living standards, people pay more and more attention to medical resources. Then, how to plan and allocate medical resources has become the focus of researchers' research. [0003] The number and management of intensive care unit (ICU) beds has always been one of the important factors reflecting medical resources. ICU length of stay and ICU death probability data also affect the planning and allocation of medical resources to a certain extent. Therefore, the prediction of ICU length of stay and ICU mortality probability has become one of the new research hotspots. [0004] At present, the prediction research of ICU length of stay and ICU mortality...

AI Technical Summary

Problems solved by technology

However, most existing methods perform poorly due to the positive skew of the data distribution of length of stay and missing data.

At the same time, some researchers have proposed some methods based on temporal convolutional neural network (TCN), but these methods ignore the relationship between different clinical characteristics when modeling patient electronic medical record data, which makes the prediction of the existing technology Less accurate and less effective

Images