Warfarin dose predictive modeling based on genetic programming and evolution strategy
ZHANG Yuzhen
TAO Yanyun
XIE Cheng
XUE Ling
ZHANG Qiyin
JIANG Bin
Abstract：To develop a method based on genetic programming (GP) and evolution strategy (ES), denoted by GPES, and to im-prove the accuracy of Warfarin dose predictive model.247 Chinese Han patients were included.Utilizing GP to evolve complex-fea-tured functions, ES evolving model coefficients and random real number of GP function,we generated prediction model.Then we com-pared our model with a linear regression model,the model developed by the International Warfarin Pharmacogenetics Consortium (IW-PC), as well as three machine learning algorithm.Among all the models,GPES got the best mean square error(MSE)(1.68×10-2) and the percentage of patients whose predicted dose of Warfarin were within ±20% of the actual dose (20%-p) (53.33%), the second best squared correlation coefficient (R2) (69.45%).Besides, GPES got the smallest absolute values of the differences be-tween MSE and 20%-p in training set and in test set, and the second smallest absolute values of the differences between R 2, respec-tively, that was δMSE(0.43×10-2), δ20%-p (0.92%) and δR2(-10.64%).GPES in this study improves the correlation, ac-curacy, applicability and extensiveness of Warfarin dose predictive model.
Keywords：WarfarinPrecision medicineDose prediction modelMachine learningEvolutionary algorithm
Publication Date：2018-01-01
Online Publishing Date：2026-09-11(First online date of this platform, not the publication date of the document)
Pages：6( 182-186,194 )