关键词: ICU后综合征；BP神经网络；影响因素；预测模型

Abstract: Objective: To construct the prediction model of Back Propagation (BP) neural network-based Post-Intensive Care Syndrome (PICS). Methods: A total of 709 patients transferred from a comprehensive ICU in a tertiary grade A hospital in Guizhou province were investigated from August 2019 to July 2020. The prediction model was constructed based on general characteristics, clinical history data and the evaluation of PICS, 80% of the participants were randomly selected as the training set, the remaining 20% participants were used as the test set, and BP algorithm was used to construct prediction model. Results: The classification accuracy of the model reached 88%, the predicted output and the expected output showed a good degree of coincidence, the overall accuracy of the model was high and the fitting effect was satisfied. Conclusion: The BP neural network-based prediction model can be used to predict the occurrence of PICS.

Key words: Post-Intensive Care syndrome; BP neural network; influencing factors; prediction model