Development and validation of a risk assessment model for seizure recurrence during anti-seizure medications withdrawal

Abstract This study is mainly to develop a risk assessment model for seizure recurrence during anti-seizure medications(ASMs) withdrawal in children and to evaluate and validate the model. Retrospective collection of children with epilepsy from January 2018 to December 2020 in the Department of Pediatrics Neurology in Binzhou Medical Universtiy Hospital who were in the process of ASMs withdrawal,They included 142 participants as the training group. The training group was divided into a relapse group and a non-relapse group according to whether the seizures had recurred,Clinical data of the training group were collected and univariate analysis was performed, then multivariate logistic regression was performed on the selected data,The prediction model was presented in the form of nomogram to evaluate the differentiation and calibration degree of the prediction model.Internal validation was also performed. The results showed that the seizure recurrence rates in the training group and the validation group were respectively 28.46% and 25%.By multivariate logistic regression, we found that seizure types, long-term freedom before ASMs withdrawal, abnormal electroencephalogram (EEG) before ASMs withdrawal and polypharmacy were independent risk factors.It is showed in the form of a formula and a nomogram.The model has good discrimination and calibration.So we conclude that seizure types, long-term seizure freedom before ASMs withdrawal, abnormal EEG before ASMs withdrawal and polypharmacy are independent risk factors.Based on these four risk factors to build a prediction model can accurately predict the risk probability of seizure recurrence, in order to reduce the ASMs in the early detection of seizure recurrence risk, advance intervention measures.