Application of a Backpropagation Artificial Neural Network in Predicting Plasma Concentration and Pharmacokinetic Parameters of Oral Single‐Dose Rosuvastatin in Healthy Subjects. Issue 7 (26th May 2020)

• Record Type: Journal Article
• Authors: Xu, Yichao; Lou, Honggang; Chen, Jinliang; Jiang, Bo; Yang, Dandan; Hu, Yin; Ruan, Zourong
• Is Part Of: Clinical pharmacology in drug development. Volume 9:Issue 7(2020)
• Page Start: 867
• Page End: 875
• Publication Date: 2020-05-26
• Subjects: backpropagation artificial neural network -- healthy subjects -- pharmacokinetic parameters -- plasma concentration -- rosuvastatin; Drugs -- Testing -- Periodicals; Drug development -- Periodicals; Clinical pharmacology -- Periodicals; 615.580724
• DOI: 10.1002/cpdd.809
• Languages: English
• ISSNs: 2160-7648

Abstract

Abstract: A backpropagation artificial neural network (BPANN) model was established for the prediction of the plasma concentration and pharmacokinetic parameters of rosuvastatin (RVST) in healthy subjects. The data (demographic characteristics and results of clinical laboratory tests) were collected from 4 bioequivalence studies using reference 10‐mg RVST calcium tablets. After the data were cleaned using extreme gradient boosting, 13 important factors were extracted to construct the BPANN model. The model was fully validated, and mean impact values (MIVs) were calculated. The model was used to predict the plasma concentration and pharmacokinetic parameters of oral single‐dose RVST in healthy subjects under fasting and fed conditions. The predicted and measured values were compared in order to evaluate the accuracy of prediction. The constructed model performed well in validation. The top 3 factors ranked by MIV related to RVST concentration are fasting/fed, time, and creatinine clearance. The time‐concentration profiles of the measured and predicted data agreed well. There were no significant differences ( P > .05) in the area under the concentration‐time curve from 0 to the last measurable concentration (AUC0‐t ) and extrapolated to infinity (AUC0‐∞ ), half‐time of elimination, peak concentration, and time to peak concentration of the measured data and data predicted by BPANN. The BPANN model has an accurate prediction ability and can be used to predict RVST concentration and pharmacokinetic parameters in healthy subjects.