Prediction model of daily living activity in convalescent patients with traumatic brain injury
Objective:To analyze the factors influencing the ADL ability and the improvement degree of ADL ability at discharge of hospitalized patients with TBI and to establish prediction models.Method:Data of 220 patients with traumatic brain injury hospitalized in the department of neurorehabilitation,Beijing Boai hospital from september 2017 to september 2020 were retrospectively collected.They were ran-domly divided into training set and validation set according to 3:1,and descriptive analysis was conducted for all variables.BI score at discharge and BI score difference between admission and discharge were used as out-come indexes.Firstly,univariate analysis was used to screen the statistically significant influencing factors,and then multivariate logistic regression analysis was used to establish the prediction model.Receiver operating characteristic(ROC)curve,calibration curve and Hosmer-Lemeshow test were used to evaluate the differentia-tion and calibration of the model in the training set and validation set.Result:Two prediction models were established by multivariate logistic regression:①Duration of disease at ad-mission,FMB at admission and BI at admission were significant influencing factors of BI at discharge.The ar-ea under the curves of the prediction model with BI at discharge was 0.957(95％CI:0.930-0.983)and 0.917(95％CI:0.839-0.994)in the training set and validation set,respectively,showing good model differentiation.Hosmer-lemeshow test results were P=0.196 and P=0.551,respectively,indicating a good calibration degree of the model.The sensitivity,specificity,and Youden index of the logistic regression model were 91.4％(95％CI:0.833-0.959)、83.3％(95％CI:0.723-0.907)and 0.747,respectively.②The duration of disease,length of stay and age at admission significantly affected the BI difference.The areas of the established prediction mod-el under the curves of training set and validation set were 0.773(95％CI:0.702-0.844)and 0.747(95％CI:0.613-0.881),and Hosmer-Lemeshow test results were P=0.721 and P=0.274,respectively,indicating good model differentiation and calibration.The sensitivity,specificity,and Youden index of the logistic regression model were 77.2％(95％CI:0.670-0.850)、64.4％(95％CI:0.522-0.750)and 0.416,respectively.Conclusion:The established two prediction models will help rehabilitation doctors to preliminarily judge the level of functional independence at discharge and the degree of functional improvement during hospitalization according to the status of patients with TBI at admission,and provide reference for rehabilitation medical work.
traumatic brain injuryactivity of daily livinglogistic regressionprediction model
NETL
NSTL
万方数据
