Bioinformatics of warfarin personalized medicine for cardiac surgery patients

Abstract

Warfarin has a large individual variability and many factors contribute to its variability. Mathematical models can quantify the impact of the factors. This study explored the effect of intravenous antibiotics on gut microbiota and menaquinone biosynthesis, investigated the effects of vitamin K concentration, gut microbiota on warfarin and constructed an information fusion perturbation theory and machine learning model (IFPTML) of warfarin blood levels, and explored the impact of gut microbiota and pharmacometabolomic data on warfarin variability based on an IFPTML model. Serum and fecal samples from 246 patients were collected to detect warfarin and vitamin K (VK1 and menaquinone-4 (MK4)) concentrations and gut microbiota diversity, respectively. The patient¿s medical records were recorded from electronic medical records. The V3¿V4 hypervariable region of the bacterial 16S rRNA gene was amplified and sequenced on a MiSeq PE300. The gut microbiota diversity of samples was analyzed in terms of ¿- and ß-diversity at the OTU level. PICRUSt2 was used for preliminary prediction of the gut microbiota function for menaquinone biosynthesis. All the continuous variables were processed by calculating the distance of the raw data with the moving average (MA ¿vki(sj)) and Euclidean distance (ED ¿¿vk(sj)¿) in different attribute groups of categorical variables. Regression and classification analyses were performed on the raw data, MA ¿vki(sj), and ED ¿¿vk(sj)¿ levels. Machine learning algorithms were chosen in the STATISTICA and WEKA software. The composition of the gut microbiota had a significant change post-intravenous antibiotics treatment. The PKPD model predicted ideal values of 62.7% for S-warfarin, 70.4% for R-warfarin, and 76.4% for INR. Prevotella and Eubacterium were identified as the main bacteria associated with warfarin. The random forest (RF) algorithm was the best algorithm for predicting continuous outputs using the raw data. General discriminant analysis (GDA) was the best algorithm for predicting the categorical outputs using the MA ¿vki(sj) data. This study highlights the importance of considering vitamin K concentration and gut microbiota when prescribing warfarin. The findings may have significant implications for the personalized use of warfarin. The RF algorithm could be used to predict the target INR, and the GDA algorithm could be used to predict the probability of being within the target INR range. The covariates of gut microbiota and pharmacometabolomic didn¿t improve the predicted performance of IFPTML model.