Mixture drug‐count response model for the high‐dimensional drug combinatory effect on myopathy

Abstract

Drug-drug interactions (DDIs) are a common cause of adverse drug events (ADEs). The electronic medical record (EMR) database and the FDA's adverse event reporting system (FAERS) database are the major data sources for mining and testing the ADE associated DDI signals. Most DDI data mining methods focus on pair-wise drug interactions, and methods to detect high-dimensional DDIs in medical databases are lacking. In this paper, we propose 2 novel mixture drug-count response models for detecting high-dimensional drug combinations that induce myopathy. The "count" indicates the number of drugs in a combination. One model is called fixed probability mixture drug-count response model with a maximum risk threshold (FMDRM-MRT). The other model is called count-dependent probability mixture drug-count response model with a maximum risk threshold (CMDRM-MRT), in which the mixture probability is count dependent. Compared with the previous mixture drug-count response model (MDRM) developed by our group, these 2 new models show a better likelihood in detecting high-dimensional drug combinatory effects on myopathy. CMDRM-MRT identified and validated (54; 374; 637; 442; 131) 2-way to 6-way drug interactions, respectively, which induce myopathy in both EMR and FAERS databases. We further demonstrate FAERS data capture much higher maximum myopathy risk than EMR data do. The consistency of 2 mixture models' parameters and local false discovery rate estimates are evaluated through statistical simulation studies.

A super‐combo‐drug test to detect adverse drug events and drug interactions from electronic health records in the era of polypharmacy

“…In our recent publications, we characterized a nonlinear drug‐count response relationship between the number of medications and an AE . In that model, however, we have not considered the weighted average among correlated drugs, nor the dosage or the drug exposure time, when testing their associations with the outcome AE.…”

“…11,25 In our recent publications, we characterized a nonlinear drug-count response relationship between the number of medications and an AE. 19,20 In that model, however, we have not considered the weighted average among correlated drugs, nor the dosage or the drug exposure time, when testing their associations with the outcome AE. We think this can be an important direction to further expand our model in order to characterize more sophisticated drug combination effects.…”

“…The methods to test single‐drug effect include disproportionality analysis, multiitem Gamma Poisson Shrinker, the likelihood ratio test, and three‐component mixture model . Other methods, such as omega shrinkage estimator, MLR, mixture dose‐response model, and graphic model, are to detect drug combination effects. In order to control confounding variable, stratification is used for unadjusted analysis .…”

Propensity score‐adjusted three‐component mixture model for drug‐drug interaction data mining in FDA Adverse Event Reporting System

“…Although PS‐adjusted analysis is clearly supreme to the unadjusted analysis in FAERS data analysis, there are other situations that unadjusted analysis is probably better. For example, if the pharmaco‐epidemiological study is designed from a longitudinal cohort, such as the electronic health record (EHR) dataset, the confounding variables, such as demographics, comedications, or comorbidity can be matched and adjusted through nested case‐control study design . In this case, in analyzing DDI‐ADE associations, the unadjusted analysis is a proper choice.…”

Deep Learning for High-Order Drug-Drug Interaction Prediction

“…The third category of methods leverage healthcare information on social media and online communities to detect DDIs that have been mentioned/inferred in online discussions and posts [16,32,39]. The last category of methods predict the probability of ADR event counts due to high-order DDIs [6,36] and use either electronic medical records or pharmacokinetic modeling to validate potential DDIs. A notable shortcoming of these methods is that they work for low-order or fixed-order DDIs but do not scale well to arbitrary orders.…”