Home    中文  
 
  • Search
  • lucene Search
  • Citation
  • Fig/Tab
  • Adv Search
Just Accepted  |  Current Issue  |  Archive  |  Featured Articles  |  Most Read  |  Most Download  |  Most Cited

Chinese Journal of Neurotraumatic Surgery(Electronic Edition) ›› 2024, Vol. 10 ›› Issue (06): 359-366. doi: 10.3877/cma.j.issn.2095-9141.2024.06.007

• Clinical Researches • Previous Articles     Next Articles

Risk prediction model of encephalocele after severe traumatic brain injury was constructed based on random forest algorithm

Xia Wang1, Yulei Liang1, Jie Shen1, Jie Yuan1, Yingpeng Wu1, Lili Zhang1, Xiangdi Chen1,()   

  1. 1. Department of Rehabilitation, the 901st Hospital of the PLA Joint Logistic Support Force,Hefei 230000,China
  • Received:2024-06-27 Online:2024-12-15 Published:2025-03-11
  • Contact: Xiangdi Chen

Abstract:

Objective

To analyze the risk factors of postoperative encephalocele in patients with severe traumatic brain injury (sTBI), and to construct a risk prediction model for postoperative encephalocele using random forest algorithm.

Methods

One hundred and eighty patients with sTBI who underwent surgical treatment from March 2020 to June 2023 in Rehabilitation Department of the 901st Hospital of the PLA Joint Logistic Support Force were retrospectively collected as the research objects and used as the modeling group. The patients were divided into the encephalocele group and the nonencephalocele group according to whether they developed encephalocele after surgery. Multivariate Logistic regression was used to analyze the risk factors of postoperative encephalocele in patients with sTBI, and R software was used to establish a random forest model to predict the occurrence of postoperative encephalocele in patients with sTBI. At the same time, 78 patients with sTBI admitted to our hospital from July 2023 to April 2024 were collected as the verification group to verify the accuracy of the random forest model.

Results

Among 180 patients with sTBI, 104 patients with sTBI had postoperative encephalocele (encephalocele group), and 76 patients with sTBI did not have postoperative encephalocele (non-encephalocele group), and the incidence of postoperative encephalocele was 57.78%(104/180). There were significant differences in preoperative intracranial pressure, preoperative GCS score, time from injury to operation, no artificial dural relaxation suture, use of elastic swimming cap and delayed hemorrhage between the encephalocele group and the non - encephalocele group (P<0.05).Multivariate Logistic regression analysis showed that preoperative intracranial pressure >60 mmHg,preoperative GCS score 3-5 points,time from injury to operation >6 h,no artificial dural relaxation suture,no use of elastic swimming cap, and delayed hemorrhage were all independent risk factors for postoperative encephalocele in sTBI patients (P<0.05). The results of random forest analysis showed that the relative important predictors of postoperative encephalocele in sTBI patients were, in order,preoperative GCS score, time from injury to operation, no artificial dural relaxation suture, preoperative intracranial pressure, delayed hemorrhage, and use of elastic swimming cap. Based on the above factors,the AUC value of the random forest model for postoperative encephalocele in sTBI patients was 0.850(sensitivity 78.85%,specificity 78.95%),which was higher than that of the multivariate Logistic regression model (AUC=0.811), and the difference was statistically significant (Z=2.296, P=0.022). Further external verification of the model found that, The AUC value, specificity and sensitivity of the verification group were 0.838, 81.40% and 74.36%, respectively, which were close to the AUC of the modeling group, with medium prediction efficiency (Z=0.740,P=0.460).

Conclusion

The influencing factors of postoperative encephalocele in sTBI patients were in order of preoperative GCS score, time from injury to operation, no artificial dural relaxation suture,preoperative intracranial pressure,delayed hemorrhage,and use of elastic swimming cap.The random forest model constructed in this study has a good risk prediction efficiency.

Key words: Severe traumatic brain injury, Postoperative encephalocele, Influencing factors, Risk prediction, Random forest

京ICP 备07035254号-20
Copyright © Chinese Journal of Neurotraumatic Surgery(Electronic Edition), All Rights Reserved.
Tel: 010-64229160 E-mail: zhsjcswk2015@163.com
Powered by Beijing Magtech Co. Ltd