Machine learning liver-injuring drug interactions with non-steroidal anti-inflammatory drugs (NSAIDs) from a retrospective electronic health record (EHR) cohort

Drug-drug interactions account for up to 30% of adverse drug reactions. Increasing prevalence of electronic health records (EHRs) offers a unique opportunity to build machine learning algorithms to identify drug-drug interactions that drive adverse events. In this study, we investigated hospitalizations' data to study drug interactions with non-steroidal anti-inflammatory drugs (NSAIDS) that result in drug-induced liver injury (DILI). We propose a logistic regression based machine learning algorithm that unearths several known interactions from an EHR dataset of about 400,000 hospitalization. Our proposed modeling framework is successful in detecting 87.5% of the positive controls, which are defined by drugs known to interact with diclofenac causing an increased risk of DILI, and correctly ranks aggregate risk of DILI for eight commonly prescribed NSAIDs. We found that our modeling framework is particularly successful in inferring associations of drug-drug interactions from relatively small EHR datasets. Furthermore, we have identified a novel and potentially hepatotoxic interaction that might occur during concomitant use of meloxicam and esomeprazole, which are commonly prescribed together to allay NSAID-induced gastrointestinal (GI) bleeding. Empirically, we validate our approach against prior methods for signal detection on EHR datasets, in which our proposed approach outperforms all the compared methods across most metrics, such as area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC). Author summary Drug-drug interactions account for nearly one-third of adverse drug reactions. Concomitant application of multiple drugs, often used to enhance therapeutic effect and selectivity, can lead to adverse drug reactions of high clinical significance. Commonly used approaches to detect adverse drug-drug interactions often rely on adverse event reporting systems and spontaneous reports. These datasets only contain specific cases where severe reactions were identified and reported by clinicians thereby overlooking a vast majority of unexpected adverse interactions that are often under-reported. The increasing prevalence of electronic health records (EHRs) provides a unique opportunity to mine these datasets and identify previous known and potentially unknown adverse drug interactions. Our proposed logistic regression-based approach identified known and unknown drug-drug interactions from relatively small EHR datasets of about 400,000 hospitalizations. Via performance comparison, our method generalizes better to analysis of EHRs when compared to other common methods currently in use by the U.S. Food & Drug Administration. Our analyses, using this proposed approach, identified 87.5% of positive controls, which are drugs that interact with diclofenac causing an increase in risk for drug-induced liver injury, and identified a novel, potentially hepatotoxic interaction between meloxicam and esomeprazole, which are commonly prescribed together to alleviate NSAID-induced gastrointestinal bleeding.

Automated summary

Drug-drug interactions contribute significantly to adverse drug reactions, with the increasing prevalence of EHRs presenting a unique opportunity to identify these interactions using machine learning algorithms. A logistic regression-based approach successfully detected known and unknown drug interactions from a relatively small EHR dataset, performing better than other common methods in predicting risk of liver injury. This study highlights the importance and effectiveness of utilizing EHR data for detecting drug-drug interactions and potential adverse events.