Establishment and validation of a risk prediction model for nonattainment plasma concentration of zonisamide in children with epilepsy
- VernacularTitle:癫痫患儿唑尼沙胺血药浓度不达标风险预测模型的建立与验证
- Author:
Huiying LI
1
;
Yan ZHANG
2
;
Fashuang LI
1
;
Linbo LI
1
;
Lilin ZHANG
1
;
Danyang REN
1
;
Caixia TU
1
;
Rundong XU
1
Author Information
1. Dept. of Pharmacy，Kunming Children’s Hospital/Children’s Hospital Affiliated to Kunming Medical University，Kunming 650228，China
2. Yunnan Institute for Food and Drug Control，Kunming 650106，China
- Publication Type:Journal Article
- Keywords:
zonisamide;
Epilepsy;
Children;
Plasma concentration;
nomogram risk prediction model;
Influencing factors
- From:
China Pharmacy
2026;37(16):2195-2201
- CountryChina
- Language:Chinese
-
Abstract:
OBJECTIVE To construct a risk prediction model for nonattainment plasma concentration of zonisamide in children with epilepsy， and to provide evidence for individualized dosing.METHODS Children with epilepsy who received standardized zonisamide treatment orally in Kunming Children’s Hospital were retrospectively selected as research subjects. Those treated from May 2022 to May 2024 were included in the modeling cohort， and those treated from June 2024 to January 2025 were included in the external validation cohort. According to plasma concentration， children were divided into therapeutic group and nonattainment group. Independent influencing factors for nonattainment plasma concentration of zonisamide were screened by collinearity diagnosis， univariate analysis， Lasso regression and random forest， based on which a nomogram risk prediction model was constructed and validated. Finally， the Shapley additive explanations （SHAP） method was used to interpret the features of the nomogram risk prediction model.RESULTS A total of 945 children with epilepsy were included in this study. There was no statistically significant difference in baseline data between the modeling cohort （531 cases） and the external validation cohort （414 cases） （ P ＞0.05）， indicating comparability. Through multi-stage feature screening， seven influencing factors were finally included to construct the prediction model， namely gender， age， dosage， aspartate transferase， total bilirubin， uric acid and vitamin D. The corrected area under the curve of the model was 0.799 in the modeling cohort and 0.762 in the external validation cohort， showing good calibration and clinical applicability. SHAP analysis showed that gender， dosage， total bilirubin， age and aspartate transferase were the top five important variables affecting model prediction.CONCLUSIONS The nomogram risk prediction model for nonattainment plasma concentration of zonisamide constructed in this study has stable predictive performance. Clinically， the risk of nonattainment plasma concentration can be quickly assessed by combining the child’s gender， age， dosage， liver and kidney function， and vitamin D， and the dosing regimen can be adjusted in time.
