Article

Machine learning-based prediction of drug-drug interactions by integrating drug phenotypic, therapeutic, chemical, and genomic properties

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Drug-drug interactions (DDIs) are an important consideration in both drug development and clinical application, especially for co-administered medications. While it is necessary to identify all possible DDIs during clinical trials, DDIs are frequently reported after the drugs are approved for clinical use, and they are a common cause of adverse drug reactions (ADR) and increasing healthcare costs. Computational prediction may assist in identifying potential DDIs during clinical trials. Here we propose a heterogeneous network-assisted inference (HNAI) framework to assist with the prediction of DDIs. First, we constructed a comprehensive DDI network that contained 6946 unique DDI pairs connecting 721 approved drugs based on DrugBank data. Next, we calculated drug-drug pair similarities using four features: phenotypic similarity based on a comprehensive drug-ADR network, therapeutic similarity based on the drug Anatomical Therapeutic Chemical classification system, chemical structural similarity from SMILES data, and genomic similarity based on a large drug-target interaction network built using the DrugBank and Therapeutic Target Database. Finally, we applied five predictive models in the HNAI framework: naive Bayes, decision tree, k-nearest neighbor, logistic regression, and support vector machine, respectively. The area under the receiver operating characteristic curve of the HNAI models is 0.67 as evaluated using fivefold cross-validation. Using antipsychotic drugs as an example, several HNAI-predicted DDIs that involve weight gain and cytochrome P450 inhibition were supported by literature resources. Through machine learning-based integration of drug phenotypic, therapeutic, structural, and genomic similarities, we demonstrated that HNAI is promising for uncovering DDIs in drug development and postmarketing surveillance.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... The models transform the DDI prediction task that infers whether or not a drug interacts with another into a binary classification problem. These methods are usually implemented according to established classifiers (e.g., KNN [16], SVM [16], logistic regression [14,20], decision tree [21], and naïve Bayes [21]), network propagation of reasoning behind drug-drug network structures [20,22], label propagation [23], random walk [15] and probabilistic soft logic [19,21], or matrix factorization [17,18,24]. Generally, traditional machine learning methods rely heavily on the quality of handcrafted features derived from the drug properties. ...
... The models transform the DDI prediction task that infers whether or not a drug interacts with another into a binary classification problem. These methods are usually implemented according to established classifiers (e.g., KNN [16], SVM [16], logistic regression [14,20], decision tree [21], and naïve Bayes [21]), network propagation of reasoning behind drug-drug network structures [20,22], label propagation [23], random walk [15] and probabilistic soft logic [19,21], or matrix factorization [17,18,24]. Generally, traditional machine learning methods rely heavily on the quality of handcrafted features derived from the drug properties. ...
... The models transform the DDI prediction task that infers whether or not a drug interacts with another into a binary classification problem. These methods are usually implemented according to established classifiers (e.g., KNN [16], SVM [16], logistic regression [14,20], decision tree [21], and naïve Bayes [21]), network propagation of reasoning behind drug-drug network structures [20,22], label propagation [23], random walk [15] and probabilistic soft logic [19,21], or matrix factorization [17,18,24]. Generally, traditional machine learning methods rely heavily on the quality of handcrafted features derived from the drug properties. ...
Article
Full-text available
The treatment of complex diseases by using multiple drugs has become popular. However, drug-drug interactions (DDI) may give rise to the risk of unanticipated adverse effects and even unknown toxicity. Therefore, for polypharmacy safety it is crucial to identify DDIs and explore their underlying mechanisms. The detection of DDI in the wet lab is expensive and time-consuming, due to the need for experimental research over a large volume of drug combinations. Although many computational methods have been developed to predict DDIs, most of these are incapable of predicting potential DDIs between drugs within the DDI network and new drugs from outside the DDI network. In addition, they are not designed to explore the underlying mechanisms of DDIs and lack interpretative capacity. Thus, here we propose a novel method of GNN-DDI to predict potential DDIs by constructing a five-layer graph attention network to identify k-hops low-dimensional feature representations for each drug from its chemical molecular graph, concatenating all identified features of each drug pair, and inputting them into a MLP predictor to obtain the final DDI prediction score. The experimental results demonstrate that our GNN-DDI is suitable for each of two DDI predicting scenarios, namely the potential DDIs among known drugs in the DDI network and those between drugs within the DDI network and new drugs from outside DDI network. The case study indicates that our method can explore the specific drug substructures that lead to the potential DDIs, which helps to improve interpretability and discover the underlying interaction mechanisms of drug pairs.
... However, wetlab experiments for verifying DDIs can drain researchers' time and resources and make it difficult for numerous and regular adoptions. Therefore, artificial intelligence (AI) models have been applied to predict DDIs [6][7][8][9]. These models have been continuously studied and improved along with the expansion and completeness of drug-database resources to support clinical decisions. ...
... These SMILES structural representations of drugs are post-processed to capture features of drug pairs associated with DDIs events [45]. Moreover, pharmacological properties such as targets [8,46], enzymes, transporters, genes and pro- teins [6,47], interaction pathways like enzymes and transporters [48][49][50][51][52][53][54][55][56][57][58][59][60][61] can also be manipulated to represent drugs features through a set of descriptors. Network interaction mining [62][63][64] and molecular graph representations have also been used to describe substructures of drugs that come in distinctive shapes and sizes or the structural relations between entities [65][66][67][68]. ...
... The classification feature constructing step usually requires the similarity analysis of paired drugs. In most studies, the chemical structural similarity was measured using the structures of the compound of drugs on DrugBank represented by their SMILES [6]. Structural representation of the drugs can be constructed using different molecular fingerprints generation techniques. ...
Article
Full-text available
Over the past decade, polypharmacy instances have been common in multi-diseases treatment. However, unwanted drug-drug interactions (DDIs) that might cause unexpected adverse drug events (ADEs) in multiple regimens therapy remain a significant issue. Since artificial intelligence (AI) is ubiquitous today, many AI prediction models have been developed to predict DDIs to support clinicians in pharmacotherapy-related decisions. However, even though DDI prediction models have great potential for assisting physicians in polypharmacy decisions, there are still concerns regarding the reliability of AI models due to their black-box nature. Building AI models with explainable mechanisms can augment their transparency to address the above issue. Explainable AI (XAI) promotes safety and clarity by showing how decisions are made in AI models, especially in critical tasks like DDI predictions. In this review, a comprehensive overview of AI-based DDI prediction, including the publicly available source for AI-DDIs studies, the methods used in data manipulation and feature preprocessing, the XAI mechanisms to promote trust of AI, especially for critical tasks as DDIs prediction, the modeling methods, is provided. Limitations and the future directions of XAI in DDIs are also discussed.
... The third category of methods, i.e., machine learning methods, has been widely used to infer drug-drug interactions [17][18][19][20][21][22][23][24][25] . Most of these methods focus on improving the performance of drug-drug interactions prediction via data integration. ...
... Dhami et al. 17 attempt to combine multiple similarity metrics (e.g., molecular feature similarity, string similarity, molecular fingerprint similarity, molecular access system) from the sole data of drug SMILES representation. The other methods [18][19][20][21][22][23][24][25] all combine multiple data sources. Data integration often combines diverse feature information such as drug adverse drug reactions (ADR) [18][19][20]23,24 , target similarity [18][19][20][22][23][24] , PPI networks 23,24 , signaling pathways 19 and so on. ...
... The other methods [18][19][20][21][22][23][24][25] all combine multiple data sources. Data integration often combines diverse feature information such as drug adverse drug reactions (ADR) [18][19][20]23,24 , target similarity [18][19][20][22][23][24] , PPI networks 23,24 , signaling pathways 19 and so on. Among these features, the information of drug chemical structures in the form of SMILES descriptors is most frequently used [17][18][19][20][21][22][23][24] . ...
Article
Full-text available
Understanding drug–drug interactions is an essential step to reduce the risk of adverse drug events before clinical drug co-prescription. Existing methods, commonly integrating heterogeneous data to increase model performance, often suffer from a high model complexity, As such, how to elucidate the molecular mechanisms underlying drug–drug interactions while preserving rational biological interpretability is a challenging task in computational modeling for drug discovery. In this study, we attempt to investigate drug–drug interactions via the associations between genes that two drugs target. For this purpose, we propose a simple f drug target profile representation to depict drugs and drug pairs, from which an l2-regularized logistic regression model is built to predict drug–drug interactions. Furthermore, we define several statistical metrics in the context of human protein–protein interaction networks and signaling pathways to measure the interaction intensity, interaction efficacy and action range between two drugs. Large-scale empirical studies including both cross validation and independent test show that the proposed drug target profiles-based machine learning framework outperforms existing data integration-based methods. The proposed statistical metrics show that two drugs easily interact in the cases that they target common genes; or their target genes connect via short paths in protein–protein interaction networks; or their target genes are located at signaling pathways that have cross-talks. The unravelled mechanisms could provide biological insights into potential adverse drug reactions of co-prescribed drugs.
... However, previously developed ligand-based and structure-based computational methods did not consider two substances in pairs simultaneously. Direct DDIs estimation methods for the pairs of substances include structure resemblance and functional similarities methods and literature-based DDIs prediction methods [9][10][11][12][13][14]. These methods deal with the pairs of substances but require information about the pharmacokinetics and pharmacodynamics [9,14], interaction profile, target and side-effects [10,13], and the phenotypic, therapeutic, chemical, and genomic properties [11] of substances or medical records [12]. ...
... Direct DDIs estimation methods for the pairs of substances include structure resemblance and functional similarities methods and literature-based DDIs prediction methods [9][10][11][12][13][14]. These methods deal with the pairs of substances but require information about the pharmacokinetics and pharmacodynamics [9,14], interaction profile, target and side-effects [10,13], and the phenotypic, therapeutic, chemical, and genomic properties [11] of substances or medical records [12]. It is clear that for new, not-yet-synthesized, and virtual substances, such information does not exist. ...
... It is clear that for new, not-yet-synthesized, and virtual substances, such information does not exist. The results of predictions of this group of methods [9][10][11][12][13][14] have often been presented as data sets containing a bulk conglomerate of information about potential DDIs predicted between the existing drugs. Such examples include 430,128 [10], 145,068 [13], and over 250,000 [14] records of unknown potential DDIs in the sets of predicted results. ...
Article
Full-text available
Drug–drug interactions (DDIs) can cause drug toxicities, reduced pharmacological effects, and adverse drug reactions. Studies aiming to determine the possible DDIs for an investigational drug are part of the drug discovery and development process and include an assessment of the DDIs potential mediated by inhibition or induction of the most important drug-metabolizing cytochrome P450 isoforms. Our study was dedicated to creating a computer model for prediction of the DDIs mediated by the seven most important P450 cytochromes: CYP1A2, CYP2B6, CYP2C19, CYP2C8, CYP2C9, CYP2D6, and CYP3A4. For the creation of structure–activity relationship (SAR) models that predict metabolism-mediated DDIs for pairs of molecules, we applied the Prediction of Activity Spectra for Substances (PASS) software and Pairs of Substances Multilevel Neighborhoods of Atoms (PoSMNA) descriptors calculated based on structural formulas. About 2500 records on DDIs mediated by these cytochromes were used as a training set. Prediction can be carried out both for known drugs and for new, not-yet-synthesized substances. The average accuracy of the prediction of DDIs mediated by various isoforms of cytochrome P450 estimated by leave-one-out cross-validation (LOO CV) procedures was about 0.92. The SAR models created are publicly available as a web resource and provide predictions of DDIs mediated by the most important cytochromes P450.
... The types of DDIs can be identified by biochemical experimental (or in vivo) methods, but experimental methods are usually time-consuming, tedious and expensive and sometimes lack reproducibility (Gao et al., 2015;Fang et al., 2017). Thus, it is highly desired to develop computational methods (or in silico) for efficiently and effectively analyzing and detecting new DDI pairs, and a variety of theoretical and computational methods have been developed to predict DDI types in recent years (Herrero-Zazo et al., 2013;Cheng and Zhao, 2014;Gottlieb et al., 2014;Zhang et al., 2015;Liu et al., 2016;Takeda et al., 2017;Zhang et al., 2017a;Zhang et al., 2017b;Andrej et al., 2018;Ryu et al., 2018;Yu et al., 2018;Lee et al., 2019;Deng et al., 2020;Feng et al., 2020;Harada et al., 2020;Lin et al., 2020;Fatehifar and Karshenas, 2021;Wang et al., 2021). Computational methods can guide experimentalists designing the best experimental scheme, narrowing the scope of candidate DDIs, and provide supporting evidence for their experimental results. ...
... Prior machine learning-based methods apply KNN (Andrej et al., 2018), SVM (Andrej et al., 2018), logistic regression (Cheng and Zhao, 2014;Gottlieb et al., 2014;Takeda et al., 2017), decision tree (Cheng and Zhao, 2014), naïve Bayes (Cheng and Zhao, 2014), and network-based label propagation (Zhang et al., 2015) and random walk (Zhang et al., 2017b) or matrix factorization (Yu et al., 2018) to detect DDIs. These methods are based on drug properties, such as chemical structure (Cheng and Zhao, 2014;Gottlieb et al., 2014;Zhang et al., 2015;Zhang et al., 2017b;Andrej et al., 2018), targets (Cheng and Zhao, 2014;Gottlieb et al., 2014;Takeda et al., 2017), Anatomical Therapeutic Chemical classification (ATC) codes (Cheng and Zhao, 2014;Gottlieb et al., 2014;Andrej et al., 2018), side effects (Gottlieb et al., 2014;Zhang et al., 2017b;Yu et al., 2018), et al. ...
... Prior machine learning-based methods apply KNN (Andrej et al., 2018), SVM (Andrej et al., 2018), logistic regression (Cheng and Zhao, 2014;Gottlieb et al., 2014;Takeda et al., 2017), decision tree (Cheng and Zhao, 2014), naïve Bayes (Cheng and Zhao, 2014), and network-based label propagation (Zhang et al., 2015) and random walk (Zhang et al., 2017b) or matrix factorization (Yu et al., 2018) to detect DDIs. These methods are based on drug properties, such as chemical structure (Cheng and Zhao, 2014;Gottlieb et al., 2014;Zhang et al., 2015;Zhang et al., 2017b;Andrej et al., 2018), targets (Cheng and Zhao, 2014;Gottlieb et al., 2014;Takeda et al., 2017), Anatomical Therapeutic Chemical classification (ATC) codes (Cheng and Zhao, 2014;Gottlieb et al., 2014;Andrej et al., 2018), side effects (Gottlieb et al., 2014;Zhang et al., 2017b;Yu et al., 2018), et al. ...
Article
Full-text available
Drug combination therapies are a promising strategy to overcome drug resistance and improve the efficacy of monotherapy in cancer, and it has been shown to lead to a decrease in dose-related toxicities. Except the synergistic reaction between drugs, some antagonistic drug–drug interactions (DDIs) exist, which is the main cause of adverse drug events. Precisely predicting the type of DDI is important for both drug development and more effective drug combination therapy applications. Recently, numerous text mining– and machine learning–based methods have been developed for predicting DDIs. All these methods implicitly utilize the feature of drugs from diverse drug-related properties. However, how to integrate these features more efficiently and improve the accuracy of classification is still a challenge. In this paper, we proposed a novel method (called NMDADNN) to predict the DDI types by integrating five drug-related heterogeneous information sources to extract the unified drug mapping features. NMDADNN first constructs the similarity networks by using the Jaccard coefficient and then implements random walk with restart algorithm and positive pointwise mutual information for extracting the topological similarities. After that, five network-based similarities are unified by using a multimodel deep autoencoder. Finally, NMDADNN implements the deep neural network (DNN) on the unified drug feature to infer the types of DDIs. In comparison with other recent state-of-the-art DNN-based methods, NMDADNN achieves the best results in terms of accuracy, area under the precision-recall curve, area under the ROC curve, F1 score, precision and recall. In addition, many of the promising types of drug–drug pairs predicted by NMDADNN are also confirmed by using the interactions checker tool. These results demonstrate the effectiveness of our NMDADNN method, indicating that NMDADNN has the great potential for predicting DDI types.
... To date, several types of computer algorithms have been adopted to design quick and reliable methods for predicting DDIs. Among them, machine learning algorithms play essential roles [8][9][10][11][12][13]. For example, Kastrin et al. [9] designed topological and semantic feature similarities, which were learnt by five classification algorithms to construct the classifier. ...
... For example, Kastrin et al. [9] designed topological and semantic feature similarities, which were learnt by five classification algorithms to construct the classifier. Cheng and Zhao [10] set up a support vector machine (SVM) model to predict DDIs, which adopted features derived from the Simplified Molecular Input Line Entry System (SMILES) and side effect similari-ties of the two drugs. Chen et al. [11] proposed a nearest neighbor algorithm-(NNA-) based model to identify DDIs, which designed a scheme to measure the similarity of two drug pairs. ...
Article
Full-text available
Combination drug therapy is an efficient way to treat complicated diseases. Drug-drug interaction (DDI) is an important research topic in this therapy as patient safety is a problem when two or more drugs are taken at the same time. Traditionally, in vitro experiments and clinical trials are common ways to determine DDIs. However, these methods cannot meet the requirements of large-scale tests. It is an alternative way to develop computational methods for predicting DDIs. Although several previous methods have been proposed, they always need several types of drug information, limiting their applications. In this study, we proposed a simple computational method to predict DDIs. In this method, drugs were represented by their fingerprint features, which are most widely used in investigating drug-related problems. These features were refined by three models, including addition, subtraction, and Hadamard models, to generate the representation of DDIs. The powerful classification algorithm, random forest, was picked up to build the classifier. The results of two types of tenfold cross-validation on the classifier indicated good performance for discovering novel DDIs among known drugs and acceptable performance for identifying DDIs between known drugs and unknown drugs or among unknown drugs. Although the classifier adopted a sample scheme to represent DDIs, it was still superior to other methods, which adopted features generated by some advanced computer algorithms. Furthermore, a user-friendly web-server, named DDIPF (http://106.14.164.77:5004/DDIPF/), was developed to implement the classifier.
... Machine learning-based methods have been widely used in the prediction of DDADRs (Yan et al. 2019;Qian et al. 2019;Liu et al. 2016;Kastrin et al. 2018;Gottlieb et al. 2012;Cheng and Zhao 2014). For example, Cheng and Zhao (2014) extracted features from simplified molecular-input line-entry system data, side effect similarities of drug pairs, and applied support vector machines (SVMs) to predict DDADRs. ...
... Machine learning-based methods have been widely used in the prediction of DDADRs (Yan et al. 2019;Qian et al. 2019;Liu et al. 2016;Kastrin et al. 2018;Gottlieb et al. 2012;Cheng and Zhao 2014). For example, Cheng and Zhao (2014) extracted features from simplified molecular-input line-entry system data, side effect similarities of drug pairs, and applied support vector machines (SVMs) to predict DDADRs. Shi et al. (2016) proposed a local classification-based model (LCM) to predict drug-drug interactions (DDIs) based on structure similarity. ...
Article
Full-text available
Concomitant drugs therapy is effective and inevitable for most patients. However, drug–drug adverse reactions (DDADRs) caused by combination drugs may bring about medical malpractices. Therefore, the accurate prediction of DDADRs is beneficial to human health and pharmaceutical researches. Currently, existing machine learning-based methods focus on a single feature. In this paper, we propose a novel method, MS-ADR, to extract multi-source features and predict DDADRs. First, we obtain four biomedical views by using four drug signed networks, respectively, namely enzyme view, indication view, side effect view, and transporter view. Then, different biomedical views are fed into graph convolutional neural networks (GCN) to extract multi-source features. Second, we propose an attention block to merge multi-source features from different biomedical views. Finally, a reconstructed drug–drug adverse reaction network is embedded to predict DDADR. The experiment shows that MS-ADR achieves better performance compared with other start-of-the-art baselines.
... However, previous works need large labeled data which may have false positive samples. And they typically either focused on the structure information or SMILES sequences (Toropov et al., 2005) of the drugs without considering the rich semantic information related to drugs (Deac et al., 2019;Huang et al., 2020;Mohamed et al., 2020;Ryu et al., 2018), or utilized knowledge graph (KG) with rich bio-medical information without considering drug molecular structure information (Karim et al., 2019;Lin et al., 2020;Zitnik et al., 2018). ...
... Vilar et al. (2012) combined the DDI and drug structural similarity matrices to generate a DDI interaction similarity matrix, and thus identifying DDI candidates. Cheng and Zhao (2014) utilized four similarities, including drug phenotypic, therapeutic, chemical structure and genomic properties, and combined with five machine learning-based models (Naive Bayes, decision tree, k-nearest neighbor, logistic regression and support vector machine) to deal with the DDI prediction task. Ryu et al. (2018) utilized the similarity of drug chemical structure as a feature and then fed the drug-drug pairs into the deep neural network (DNN) to predict the interaction type. ...
Article
Full-text available
Motivation: Adverse drug-drug interactions (DDIs) are crucial for drug research and mainly cause morbidity and mortality. Thus, the identification of potential DDIs is essential for doctors, patients, and the society. Existing traditional machine learning models rely heavily on handcraft features and lack generalization. Recently, the deep learning approaches that can automatically learn drug features from the molecular graph or drug-related network have improved the ability of computational models to predict unknown DDIs. However, previous works utilized large labeled data and merely considered the structure or sequence information of drugs without considering the relations or topological information between drug and other biomedical objects (e.g., gene, disease, and pathway), or considered knowledge graph (KG) without considering the information from the drug molecular structure. Results: Accordingly, to effectively explore the joint effect of drug molecular structure and semantic information of drugs in knowledge graph for DDI prediction, we propose a multi-scale feature fusion deep learning model named MUFFIN. MUFFIN can jointly learn the drug representation based on both the drug-self structure information and the KG with rich bio-medical information. In MUFFIN, we designed a bi-level cross strategy that includes cross- and scalar-level components to fuse multi-modal features well. MUFFIN can alleviate the restriction of limited labeled data on deep learning models by crossing the features learned from large-scale KG and drug molecular graph. We evaluated our approach on three datasets and three different tasks including binary-class, multi-class, and multi-label DDI prediction tasks. The results showed that MUFFIN outperformed other state-of-the-art baselines. Availability: The source code and data are available at https://github.com/xzenglab/MUFFIN. Supplementary information: Supplementary data are available at Bioinformatics online.
... Gottlieb et al. (2012) constructed drug feature vectors based on seven types of drug-drug similarities to describe drug-drug pairs, and then applies logistic regression model to predict DDI. A heterogeneous network-assisted inference framework (Cheng and Zhao 2014) is proposed to assist the prediction of DDI, which integrates multiple similarity features and applies five classification models. Zhang et al. (2015) built a high-order similarity weight network and used a semi-supervised label propagation algorithm to predict drug-drug interactions. ...
... Drug chemical structure information is critical and easy to obtain. Drug feature based on the chemical structure has already been successfully developed in many DDI prediction studies (Cheng and Zhao 2014;Zhang et al. 2015;Ryu et al. 2018). According to Zhang et al. (2017b), we construct the drug feature based on 881 types of chemical substructures, as known as chemical fingerprints defined in PubChem (Li et al. 2010). ...
Article
Full-text available
Drug–Drug interaction (DDI) prediction is essential in pharmaceutical research and clinical application. Existing computational methods mainly extract data from multiple resources and treat it as binary classification. However, this cannot unambiguously tell the boundary between positive and negative samples owing to the incompleteness and uncertainty of derived data. A granular computing method called three-way decision is proved to be effective in making uncertain decision, but it relies on supplementary information to make delay decision. Recently, biomedical knowledge graph has been regarded as an important source to obtain abundant supplementary information about drugs. This paper proposes a three-way decision-based method called 3WDDI, in combination with knowledge graph embedding as supplementary features to enhance DDI prediction. The drug pairs are divided into positive, negative and boundary regions by Convolutional Neural Network (CNN) according to drug chemical structure feature. Further, delay decision is made for objects in the boundary region by integrating knowledge graph embedding feature to promote the accuracy of decision-making. The empirical results show that 3WDDI yields up to 0.8922, 0.9614, 0.9582, 0.8930 for Accuracy, AUPR, AUC and F1-score, respectively, and outperforms several baseline models.
... A substantial amount of work in DDI focuses on homogeneous data types such as text [31,11], textual representation of the structural data of drugs [20,3] and genetic data [37]. Recent approaches consider phenotypic, therapeutic, structural, genomic and reactive drug properties [13] or their combinations [16] to characterize drug interactivity but this type of information only serves to extract in vivo/vitro discoveries. ...
... Most similarity-based methods for DDI discovery/prediction construct NLP-based kernels from literature data [39,15]. A different direction is to learn kernels from different types of data such as molecular and structural properties of the drugs and then using these multiple kernels to predict DDIs [13,16]. Recently embeddings have been employed for learning from a single data source [36,10]. ...
Preprint
Full-text available
Predicting and discovering drug-drug interactions (DDIs) using machine learning has been studied extensively. However, most of the approaches have focused on text data or textual representation of the drug structures. We present the first work that uses multiple data sources such as drug structure images, drug structure string representation and relational representation of drug relationships as the input. To this effect, we exploit the recent advances in deep networks to integrate these varied sources of inputs in predicting DDIs. Our empirical evaluation against several state-of-the-art methods using standalone different data types for drugs clearly demonstrate the efficacy of combining heterogeneous data in predicting DDIs.
... In the light of principles of the state-of-art ML algorithms described throughout section 5 , in this section, we present paradigms of ML studies reported in biomedical literature. Two main types of ML approaches -(network-based [108][109][110][111] and similarity-based [ 17 , 112-120 ]) -have been adopted by researchers for the purposes of predicting DDI safety signals. ...
... Numerous PD DDIs have been discovered with high accuracy score (0.82) and modest recall score (0.62). In 2014 [ 111 ], Cheng et al., adopted a heterogeneous network-assisted inference (HNAI) framework for large-scale prediction of ligand-receptor DDIs. Their work depended on integrating drug phenotypic, therapeutic, chemical, and genomic properties. ...
Article
Full-text available
The objective of this article is to review the application of informatics-driven approaches in the pharmacovigilance field with focus on drug-drug interaction (DDI) safety signal discovery using various data sources. Signal can be a new safety information or new aspect to already known adverse drug reaction which is possibly causally related to a medication/medications that warrants further investigation to accept or refute. Signals can be detected from different data sources such as spontaneous reporting system, scientific literature, biomedical databases and electronic health records. This review is substantiated based on the fact that DDIs are contributing to a public health problem represented in 6-30% adverse drug event occurrences. In this article, we review informatics-driven approaches applied by authors focusing on DDI signal detection using different data sources. The aim of this article is not to laboriously survey all PV literature. As an alternative, we discussed informatics-driven methods used to discover DDI signals and various data sources reinforced with instances of studies from PV literature. The adoption of informatics-driven approaches can complement and optimize the practice of safety signal detection. However, further researches should be carried out to evaluate the efficiency of those approaches and to address the limitations of external validation, implementation and adoption in real clinical environments and by the regulatory bodies.
... Conventional classifier-based models. Cheng et al. [45] propose a heterogeneous network-assisted inference (HNAI) framework employing five predictive models including LR, NB, DT, SVM, and k-NN to build prediction models. They use several similarities of drug pairs as the features of each drug-drug pair. ...
... Some methods using matrix factorization also obtain relatively promising results [69], [70]. In addition, some studies have shown the importance of integrating heterogeneous drug features [45], [58], [85], [135]. ...
Article
Full-text available
The detection of drug-drug interactions (DDIs) is a crucial task for drug safety surveillance, which provides effective and safe co-prescriptions of multiple drugs. Since laboratory researches are often complicated, costly and time-consuming, it's urgent to develop computational approaches to detect drug-drug interactions. In this paper, we conduct a comprehensive review of state-of-the-art computational methods falling into three categories: literature-based extraction methods, machine learning-based prediction methods and pharmacovigilance-based data mining methods. Literature-based extraction methods detect DDIs from published literature using natural language processing techniques; machine learning-based prediction methods build prediction models based on the known DDIs in databases and predict novel ones; pharmacovigilance-based data mining methods usually apply statistical techniques on various electronic data to detect drug-drug interaction signals. We first present the taxonomy of drug-drug interaction detection methods and provide the outlines of three categories of methods. Afterwards, we respectively introduce research backgrounds and data sources of three categories, and illustrate their representative approaches as well as evaluation metrics. Finally, we discuss the current challenges of existing methods and highlight potential opportunities for future directions.
... In silico methods integrating molecular data with pharmacological data could potentially identify drug combinations with some limitations [124]. A heterogeneous network-assisted inference (HNAI) framework was developed using drug-drug interaction pairs connecting approved drugs, phenotypic similarity, therapeutic similarity, chemical structure similarity, and genomic similarity using naive Bayes, decision tree, k-nearest neighbor (KNN), logistic regression, and SVM algorithms [125]. Then, the DDIGIP method, in which the Gaussian interaction profile (GIP) kernel and the regularized least squares (RLS) classifier were implemented, was based on drug-drug interactions (DDIs) [126]. ...
Article
Full-text available
Resistance to therapy remains a major cause of cancer treatment failures, resulting in many cancer-related deaths. Resistance can occur at any time during the treatment, even at the beginning. The current treatment plan is dependent mainly on cancer subtypes and the presence of genetic mutations. Evidently, the presence of a genetic mutation does not always predict the therapeutic response and can vary for different cancer subtypes. Therefore, there is an unmet need for predictive models to match a cancer patient with a specific drug or drug combination. Recent advancements in predictive models using artificial intelligence have shown great promise in preclinical settings. However, despite massive improvements in computational power, building clinically useable models remains challenging due to a lack of clinically meaningful pharmacogenomic data. In this review, we provide an overview of recent advancements in therapeutic response prediction using machine learning, which is the most widely used branch of artificial intelligence. We describe the basics of machine learning algorithms, illustrate their use, and highlight the current challenges in therapy response prediction for clinical practice.
... Gottlieb et al. present a method to predict DDI considering the structural similarity and side effects of known drug pairs [7]. Lee et al. propose a model to predict DDI using a feed-forward deep neural network that takes reduced similarity profiles generated from autoencoders as input [8]. Cheng et al. created drug-drug similarity pairs based on multiple features and applied five predictive models based on naive Bayes, decision tree, k-nearest neighbour, logistic regression, and support vector machine [9]. Zhang et al. proposed a method for finding DDI by applying a label propagation algorithm on a network of structural and side effect similarity of drugs [10]. ...
... Gottlieb et al. put forward a model named INDI, extracting feature vectors by calculating seven drug similarities and predicting interactions of the drugs by logistic regression [15]. Cheng et al. merged many drug similarities to express drug-drug pairs and exploited five classifiers to build predicting models [16]. Ferdousi et al. provided a method to construct embedding vectors of drugs, using four biological elements, including carriers, transporters, enzymes and targets (CTET), to predict potential DDIs through a Russell-Rao similarity [17]. ...
Article
Full-text available
During the development of drug and clinical applications, due to the co-administration of different drugs that have a high risk of interfering with each other’s mechanisms of action, correctly identifying potential drug–drug interactions (DDIs) is important to avoid a reduction in drug therapeutic activities and serious injuries to the organism. Therefore, to explore potential DDIs, we develop a computational method of integrating multi-level information. Firstly, the information of chemical sequence is fully captured by the Natural Language Processing (NLP) algorithm, and multiple biological function similarity information is fused by Similarity Network Fusion (SNF). Secondly, we extract deep network structure information through Hierarchical Representation Learning for Networks (HARP). Then, a highly representative comprehensive feature descriptor is constructed through the self-attention module that efficiently integrates biochemical and network features. Finally, a deep neural network (DNN) is employed to generate the prediction results. Contrasted with the previous supervision model, BioChemDDI innovatively introduced graph collapse for extracting a network structure and utilized the biochemical information during the pre-training process. The prediction results of the benchmark dataset indicate that BioChemDDI outperforms other existing models. Moreover, the case studies related to three cancer diseases, including breast cancer, hepatocellular carcinoma and malignancies, were analyzed using BioChemDDI. As a result, 24, 18 and 20 out of the top 30 predicted cancer-related drugs were confirmed by the databases. These experimental results demonstrate that BioChemDDI is a useful model to predict DDIs and can provide reliable candidates for biological experiments. The web server of BioChemDDI predictor is freely available to conduct further studies.
... The global structure similarity has been used to build connections between the off-targets 23 . In DDI, the similarities of positive DDI pairs were significantly higher than those of random DDI pairs 77 . With similarity inherence and the collective inference approach, the Probabilistic Soft Logic language 65 conveys the similarity of drugs (targets) to new targets (drugs) through the known DTIs. ...
Preprint
Full-text available
Omics and drug molecules become increasingly influential in identifying disease mechanisms and drug response. Because diseases and drug responses are co-expressed and regulated in the relevant omics interactions, the traditional way that grabbing molecular data from single isolated layers cannot always obtain valuable inference. Also, adverse effects exist in drugs that impair patients, and launching new medicines for diseases is costly. To resolve the above difficulties, systems biology is then applied to predict potential molecular interaction elements by integrating omics data from genomic, proteomic, transcriptional, and metabolic layers. Combined with known drug reactions, the resulting models improve medicines’ therapeutical performance by re-purposing the existing drugs and combining drug molecules without off-target effects. Based on the identified computational models, drug administration control laws are designed to balance toxicity and efficacy. This review introduces biomedical applications and analyses of interactions among omics and drug molecules for modeling disease mechanism and drug response. The therapeutical performance can be improved by combining the predictive and computational models with drug administration designed by control laws. The challenges are discussed for its clinic uses.
... These data include chemical structures 8,9 , target information 10 , compound-target docking scores 11 , and drug side effects 12 . There have also been efforts at predicting DDIs by integrating molecular and pharmacological data 13,14 . Another group of methods analyze medical literature and/or electronic medical records to extract potential DDIs. ...
Article
Many adverse drug reactions (ADRs) are caused by drug-drug interactions (DDIs), meaning they arise from concurrent use of multiple medications. Detecting DDIs using observational data has at least three major challenges: (1) The number of potential DDIs is astronomical; (2) Associations between drugs and ADRs may not be causal due to observed or unobserved confounding; and (3) Frequently co-prescribed drug pairs that each independently cause an ADR do not necessarily causally interact, where causal interaction means that at least some patients would only experience the ADR if they take both drugs. We address (1) through data mining algorithms pre-filtering potential interactions, and (2) and (3) by fitting causal interaction models adjusting for observed confounders and conducting sensitivity analyses for unobserved confounding. We rank candidate DDIs robust to unobserved confounding more likely to be real. Our rigorous approach produces far fewer false positives than past applications that ignored (2) and (3).
... Prediction of AEs resulting from unknown drug-drug interactions (DDI) is another prevalent research area. A heterogeneous network-assisted inference (HNAI) framework was proposed by Cheng and Zhao [20] to assist with the prediction of ligand-receptor DDIs. Drug-drug pair similarity is calculated on the basis of four features: phenotypic, therapeutic, chemical structure similarity, and genomic similarity. ...
Article
Full-text available
Proactive and objective regulatory risk management of ongoing clinical trials is limited, especially when it involves the safety of the trial. We seek to prospectively evaluate the risk of facing adverse outcomes from standardized and routinely collected protocol data. We conducted a retrospective cohort study of 2860 Phase 2 and Phase 3 trials that were started and completed between 1993 and 2017 and documented in ClinicalTrials.gov. Adverse outcomes considered in our work include Serious or Non-Serious as per the ClinicalTrials.gov definition. Random-forest-based prediction models were created to determine a trial’s risk of adverse outcomes based on protocol data that is available before the start of a trial enrollment. A trial’s risk is defined by dichotomic (classification) and continuous (log-odds) risk scores. The classification-based prediction models had an area under the curve (AUC) ranging from 0.865 to 0.971 and the continuous-score based models indicate a rank correlation of 0.6–0.66 (with p-values < 0.001), thereby demonstrating improved identification of risk of adverse outcomes. Whereas related frameworks highlight the prediction benefits of incorporating data that is highly context-specific, our results indicate that Adverse Event (AE) risks can be reliably predicted through a framework of mild data requirements. We propose three potential applications in leading regulatory remits, highlighting opportunities to support regulatory oversight and informed consent decisions.
... Gottlieb et al. [11] constructed a model named INDI which calculated seven types of similarity and used a weighted logistic regression classifier to predict drug-drug interactions. Cheng et al. [12] integrated a variety of drug-drug similarities to describe drugdrug pairs, and used five classifiers to build the prediction models. Foukoue et al. [13] proposed a framework named Tiresias which constructed a knowledge graph through semantic integration of data and utilized the graph to calculate similarities among drugs. ...
Preprint
Drug-drug interactions are one of the main concerns in drug discovery. Accurate prediction of drug-drug interactions plays a key role in increasing the efficiency of drug research and safety when multiple drugs are co-prescribed. With various data sources that describe the relationships and properties between drugs, the comprehensive approach that integrates multiple data sources would be considerably effective in making high-accuracy prediction. In this paper, we propose a Deep Attention Neural Network based Drug-Drug Interaction prediction framework, abbreviated as DANN-DDI, to predict unobserved drug-drug interactions. First, we construct multiple drug feature networks and learn drug representations from these networks using the graph embedding method; then, we concatenate the learned drug embeddings and design an attention neural network to learn representations of drug-drug pairs; finally, we adopt a deep neural network to accurately predict drug-drug interactions. The experimental results demonstrate that our model DANN-DDI has improved prediction performance compared with state-of-the-art methods. Moreover, the proposed model can predict novel drug-drug interactions and drug-drug interaction-associated events.
... Their model final predicts potential drug combinations would be the drug-drug pairs with high overlap in the disease network, affecting multiple key signaling modules. Based on the hypothesis that drug-drug pairs with more similar chemical structures, target proteins, adverse drug reactions, and therapeutic purposes have a higher probability of effective drug combinations, Cheng and Zhao (Cheng and Zhao, 2014) applied machine learning techniques to predict drug combinations. Therefore, they calculated a lot of drug-drug similarities, such as phenotypic similarity, therapeutic similarity, chemical structure similarity, and genomic similarity. ...
Preprint
Full-text available
Some studies have shown that efficacious drug combination can increase the therapeutic effect, and decrease drug toxicity and side-effects. Thus, drug combinations have been widely used in the treatment of complex diseases, especially cancer. However, experiment-based methods are extremely costly in time and money. Computational models can greatly reduce the cost, but most of the models do not use the data of more than two drugs and lose a lot of useful information. Here, we used high-order drug combination information and developed a hypergraph random walk with restart model (HRWR) for efficacious drug combination prediction. As a result, compared with the other methods by leave-one-out cross-validation (LOOCV), the Area Under Receiver Operating Characteristic Curve (AUROC) of the HRWR algorithm were higher than others. Moreover, the case studies of lung cancer, breast cancer, and colorectal cancer showed that HRWR had a powerful ability to predict potential efficacious combinations, which provides new prospects for cancer treatment. The code and dataset of HRWR are freely available at https://github.com/wangqi27/HRWR .
... Various machine learning methods have been proved as a promising method to provide a preliminary screening of DDIs for further experimental validation with the advantages of both high efficiency and low costs. Generally, the machine learning-based methods [4][5][6][7][8][9][10][11][12][13][14][15] use the approved DDIs training the predictive models to infer the potential DDIs among massive unlabeled drug pairs by extracting the drug features from diverse drug property source, such as chemical structure [4,[6][7][8][9], targets [4][5][6][7], anatomical taxonomy [5,8,10] and phenotypic observation [5,7,9,10], or extracting the drug similarity features [5,6,9,10,[16][17][18], or training the deep learning models to extract better features from raw features [19][20][21]. However, most of these existing methods focus on whether a drug interacts with another or not. ...
Preprint
Full-text available
Although the polypharmacy has both higher therapeutic efficacy and less drug resistance in combating complex diseases, drug-drug interactions (DDIs) may trigger unexpected pharmacological effects, such as side effects, adverse reactions, or even serious toxicity. Thus, it is crucial to identify DDIs and explore its underlying mechanism (e.g., DDIs types) for polypharmacy safety. However, the detection of DDIs in assays is still time-consuming and costly, due to the need of experimental search over a large drug combinational space. Machine learning methods have been proved as a promising and efficient method for preliminary DDI screening. Most shallow learning-based predictive methods focus on whether a drug interacts with another or not. Although deep learning (DL)-based predictive methods address a more realistic screening task for identifying the DDI types, they only predict the DDI types of known DDI, ignoring the structural relationship between DDI entries, and they also cannot reveal the knowledge about the dependence between DDI types. Thus, here we proposed a novel end-to-end deep learning-based predictive method (called MTDDI) to predict DDIs as well as its types, exploring the underlying mechanism of DDIs. MTDDI designs an encoder derived from enhanced deep relational graph convolutional networks to capture the structural relationship between multi-type DDI entries, and adopts the tensor-like decoder to uniformly model both single-fold interactions and multi-fold interactions to reflect the relation between DDI types. The results show that our MTDDI is superior to other state-of-the-art deep learning-based methods. For predicting the multi-type DDIs with unknown DDIs in case of both single-fold DDIs and multi-fold DDIs, we validated the effectiveness and the practical capability of our MTDDI. More importantly, MTDDI can reveal the dependency between DDI types. These crucial observations are beneficial to uncover the mechanism and regularity of DDIs.
... e computational prediction may assist in identifying potential drug-drug interactions [45]. Several AI-based models were developed for the a priori detection of drug-drug interactions from biological, chemical, and pharmacokinetic data with high accuracy in the academic setting, but none have reached the clinical implementation stage [46,47]. ...
Article
Full-text available
Advances in cancer molecular profiling have enabled the development of more effective approaches to the diagnosis and personalized treatment of tumors. However, treatment planning has become more labor intensive, requiring hours or even days of clinician effort to optimize an individual patient case in a trial-and-error manner. Lessons learned from the world cancer programs provide insights into ways to develop approaches for the treatment strategy definition which can be introduced into clinical practice. This article highlights the variety of breakthroughs in patients’ cancer treatment and some challenges that this field faces now in Russia. In this report, we consider the key characteristics for planning an optimal clinical treatment regimen and which should be included in the algorithm of clinical decision support systems. We discuss the perspectives of implementing artificial intelligence-based systems in cancer treatment planning in Russia.
... Sun et al. [20] predict drug combinations by integrating the gene expression data of multiple drugs, which enhances the performance of algorithms, indicating that gene expression is also a discriminative feature for drug combinations. To validate the role of genomic features, HNAI [21] fuses the drug phenotypic, therapeutic, structural, and genomic similarities to the prediction of drug combinations by using five machine learning-based classifiers. ...
Article
Full-text available
Background Drug combination, offering an insight into the increased therapeutic efficacy and reduced toxicity, plays an essential role in the therapy of many complex diseases. Although significant efforts have been devoted to the identification of drugs, the identification of drug combination is still a challenge. The current algorithms assume that the independence of feature selection and drug prediction procedures, which may result in an undesirable performance. Results To address this issue, we develop a novel Se mi-supervised H eterogeneous N etwork E mbedding algorithm (called SeHNE) to predict the combination patterns of drugs by exploiting the graph embedding. Specifically, the ATC similarity of drugs, drug–target, and protein–protein interaction networks are integrated to construct the heterogeneous networks. Then, SeHNE jointly learns drug features by exploiting the topological structure of heterogeneous networks and predicting drug combination. One distinct advantage of SeHNE is that features of drugs are extracted under the guidance of classification, which improves the quality of features, thereby enhancing the performance of prediction of drugs. Experimental results demonstrate that the proposed algorithm is more accurate than state-of-the-art methods on various data, implying that the joint learning is promising for the identification of drug combination. Conclusions The proposed model and algorithm provide an effective strategy for the prediction of combinatorial patterns of drugs, implying that the graph-based drug prediction is promising for the discovery of drugs.
... Gottlieb et al. [11] constructed a model named INDI which calculated seven types of similarity and used a weighted logistic regression classifier to predict drug-drug interactions. Cheng et al. [12] integrated a variety of drug-drug similarities to describe drugdrug pairs, and used five classifiers to build the prediction models. Foukoue et al. [13] proposed a framework named Tiresias which constructed a knowledge graph through semantic integration of data and utilized the graph to calculate similarities among drugs. ...
Article
Full-text available
Drug-drug interactions are one of the main concerns in drug discovery. Accurate prediction of drug-drug interactions plays a key role in increasing the efficiency of drug research and safety when multiple drugs are co-prescribed. With various data sources that describe the relationships and properties between drugs, the comprehensive approach that integrates multiple data sources would be considerably effective in making high-accuracy prediction. In this paper, we propose a Deep Attention Neural Network based Drug-Drug Interaction prediction framework, abbreviated as DANN-DDI, to predict unobserved drug-drug interactions. First, we construct multiple drug feature networks and learn drug representations from these networks using the graph embedding method; then, we concatenate the learned drug embeddings and adopt an attention neural network to learn representations of drug-drug pairs; finally, we design a deep neural network to accurately predict drug-drug interactions. The experimental results demonstrate that our model DANN-DDI has improved prediction performance compared with state-of-the-art methods. Moreover, the proposed model can predict novel drug-drug interactions and drug-drug interaction-associated events.
... This amplification of available data size encourages the application of data-driven analysis tools. Typically, techniques of machine learning have been widely adopted in the chemical industry for purposes such as de novo molecular design [270] [285], chemical properties prediction [286] [287] [288], prediction of protein structures [289] [290], synthetic route analysis [291] [292] and retro-synthesis [293]. Commonly used algorithms in these categories include support vector machines (SVM) [ [304]. ...
Thesis
The design of deep neural networks (DNNs) involves the explicit definition of network architecture as well as the training of the network weights. Each process can be formulated into an optimisation algorithm and can be investigated with regard to optimisation performance. The training of the network weights is defined as a minimisation of the objective function with regard to network parameters. The architecture search is an optimisation of the objective function with regard to the presence/absence of layers or neurons. I draw similarity between the two scenarios, and propose frameworks that define either the training or the architectural optimisation of DNNs, or a combination of both. The contribution of the thesis is six-fold, in which I propose: 1) a quasi-Newton training algorithm based on Truncated Newton and Gradient Flow methods, 2) a lifting scheme to allow network sparsification, 3) a lifting framework to automatically evolve neural architectures, 4) a multi-scale hierarchical search framework involving sensitivity analysis suitable for the training of neural networks, 5) a heuristic search algorithm for architectural optimisation of a dynamic model, and 6) a dynamic cascade learning model solved in the context of de novo drug design. In each contribution, I define the optimisation problem and solve the optimisation problem under different frameworks. The ultimate aim of this research is to facilitate the democratisation of AI, enabling people with less domain expertise to participate in the design of a deep neural network under a guided framework.
... On this basis, Gottlied et al. [18] exploited more different drug-drug similarities and proposed another logistic regression model. Two similarity-based models based on drug interaction profile fingerprints were proposed [16,19] and a heterogeneous network-assisted inference framework was introduced by Cheng et al. [20]. Some other algorithms were extended on the task of DDIs' prediction. ...
Article
Full-text available
Background Drug–drug interactions (DDIs) are the reactions between drugs. They are compartmentalized into three types: synergistic, antagonistic and no reaction. As a rapidly developing technology, predicting DDIs-associated events is getting more and more attention and application in drug development and disease diagnosis fields. In this work, we study not only whether the two drugs interact, but also specific interaction types. And we propose a learning-based method using convolution neural networks to learn feature representations and predict DDIs. Results In this paper, we proposed a novel algorithm using a CNN architecture, named CNN-DDI, to predict drug–drug interactions. First, we extract feature interactions from drug categories, targets, pathways and enzymes as feature vectors and employ the Jaccard similarity as the measurement of drugs similarity. Then, based on the representation of features, we build a new convolution neural network as the DDIs’ predictor. Conclusion The experimental results indicate that drug categories is effective as a new feature type applied to CNN-DDI method. And using multiple features is more informative and more effective than single feature. It can be concluded that CNN-DDI has more superiority than other existing algorithms on task of predicting DDIs.
... However, traditional biological experiments for DTI detection are normally costly and timeconsuming 6,7 . In the past decades, many computational approaches for DTI identification have been developed to narrow down the search space of drug and protein candidates for reducing cost and accelerating efficiency of drug discovery and development [8][9][10] . Generally, the approaches for in silico DTI prediction can be classified into three categories: structure-based approaches, ligand-based approaches, and hybrid approaches 11 . ...
Article
Full-text available
Prediction of drug-target interactions (DTI) plays a vital role in drug development in various areas, such as virtual screening, drug repurposing and identification of potential drug side effects. Despite extensive efforts have been invested in perfecting DTI prediction, existing methods still suffer from the high sparsity of DTI datasets and the cold start problem. Here, we develop KGE_NFM, a unified framework for DTI prediction by combining knowledge graph (KG) and recommendation system. This framework firstly learns a low-dimensional representation for various entities in the KG, and then integrates the multimodal information via neural factorization machine (NFM). KGE_NFM is evaluated under three realistic scenarios, and achieves accurate and robust predictions on four benchmark datasets, especially in the scenario of the cold start for proteins. Our results indicate that KGE_NFM provides valuable insight to integrate KG and recommendation system-based techniques into a unified framework for novel DTI discovery.
... Contemporary similarity-based approaches assume that similar drugs may have similar interactions. Such similarities are obtained based on drug structures, targets, ontologies, and side effects, which are utilized as features for machine-learning training [5][6][7][8][9]. In the case of networkbased approaches, novel interactions are inferred via network analyses. ...
Article
Full-text available
Adverse drug-drug interaction (DDI) is a major concern to polypharmacy due to its unexpected adverse side effects and must be identified at an early stage of drug discovery and development. Many computational methods have been proposed for this purpose, but most require specific types of information, or they have less concern in interpretation on underlying genes. We propose a deep learning-based framework for DDI prediction with drug-induced gene expression signatures so that the model can provide the expression level of interpretability for DDIs. The model engineers dynamic drug features using a gating mechanism that mimics the co-administration effects by imposing attention to genes. Also, each side-effect is projected into a latent space through translating embedding. As a result, the model achieved an AUC of 0.889 and an AUPR of 0.915 in unseen interaction prediction, which is competitively very accurate and outperforms other state-of-the-art methods. Furthermore, it can predict potential DDIs with new compounds not used in training. In conclusion, using drug-induced gene expression signatures followed by gating and translating embedding can increase DDI prediction accuracy while providing model interpretability. The source code is available on GitHub (https://github.com/GIST-CSBL/DeSIDE-DDI).
Article
Full-text available
Background Deep learning methods are a proven commodity in many fields and endeavors. One of these endeavors is predicting the presence of adverse drug–drug interactions (DDIs). The models generated can predict, with reasonable accuracy, the phenotypes arising from the drug interactions using their molecular structures. Nevertheless, this task requires improvement to be truly useful. Given the complexity of the predictive task, an extensive benchmarking on structure-based models for DDIs prediction was performed to evaluate their drawbacks and advantages. Results We rigorously tested various structure-based models that predict drug interactions using different splitting strategies to simulate different real-world scenarios. In addition to the effects of different training and testing setups on the robustness and generalizability of the models, we then explore the contribution of traditional approaches such as multitask learning and data augmentation. Conclusion Structure-based models tend to generalize poorly to unseen drugs despite their ability to identify new DDIs among drugs seen during training accurately. Indeed, they efficiently propagate information between known drugs and could be valuable for discovering new DDIs in a database. However, these models will most probably fail when exposed to unknown drugs. While multitask learning does not help in our case to solve the problem, the use of data augmentation does at least mitigate it. Therefore, researchers must be cautious of the bias of the random evaluation scheme, especially if their goal is to discover new DDIs.
Article
Small molecule transporters (SMTs) in the ABC and SLC families are important players in disposition of diverse endo‐ and xenobiotics. Interactions of environmental chemicals with these transporters were first postulated in the 1990s, and since validated in numerous in vitro and in vivo scenarios. Recent results on the co‐crystal structure of ABCB1 with the flame‐retardant BDE‐100 demonstrate that a diverse range of man‐made and natural toxic molecules, hereafter termed transporter‐interfering chemicals (TICs), can directly bind to SMTs and interfere with their function. TIC‐binding modes mimic those of substrates, inhibitors, modulators, inducers, and possibly stimulants through direct and allosteric mechanisms. Similarly, the effects could directly or indirectly agonize, antagonize or perhaps even prime the SMT system to alter transport function. Importantly, TICs are distinguished from drugs and pharmaceuticals that interact with transporters in that exposure is unintended and inherently variant. Here, we review the molecular mechanisms of environmental chemical interaction with SMTs, the methodological considerations for their evaluation, and the future directions for TIC discovery.
Chapter
Artificial intelligence (AI) as a technology concept is making a major impact on a wide range of industries and sectors. This is largely attributed to technical advancements in machine and especially deep learning methodologies fueled by improved computational capabilities which have led to sophisticated approaches in applying AI to various scenarios. These AI applications aim to improve productivity, decrease cost, and comprehend the ever-increasing volumes of data available to ultimately provide actionable insights. Included in this paradigm shift is medicine, where AI is beginning to enable clinical assistance, decision support, improved management and accelerated scientific discovery and development.
Article
It is crucial to identify DDIs and explore their underlying mechanism (e.g., DDIs types) for polypharmacy safety. However, the detection of DDIs in assays is still time-consuming and costly, due to the need for experimental search over a large space of drug combinations. Thus, many computational methods have been developed to predict DDIs, most of them focusing on whether a drug interacts with another or not. And a few deep learning-based methods address a more realistic screening task for identifying various DDI types, but they assume a DDI only triggers one pharmacological effect, while a DDI can trigger more types of pharmacological effects. Thus, here we proposed a novel end-to-end deep learning-based method (called deepMDDI) for the Multi-label prediction of Drug-Drug Interactions. deepMDDI contains an encoder derived from relational graph convolutional networks and a tensor-like decoder to uniformly model interactions. deepMDDI is not only efficient for DDI transductive prediction, but also inductive prediction. The experimental results show that our model is superior to other state-of-the-art deep learning-based methods. We also validated the power of deepMDDI in the DDIs multi-label prediction and found several new valid DDIs in the case study. In conclusion, deepMDDI is beneficial to uncover the mechanism and regularity of DDIs.
Article
Motivation Polypharmacy is the combined use of drugs for the treatment of diseases. However, it often shows a high risk of side effects. Due to unnecessary interactions of combined drugs, the side effects of polypharmacy increase the risk of disease and even lead to death. Thus, obtaining abundant and comprehensive information on the side effects of polypharmacy is a vital task in the healthcare industry. Early traditional methods employed machine learning techniques to predict side effects. However, they often make costly efforts to extract features of drugs for prediction. Later, several methods based on knowledge graphs are proposed. They are reported to outperform traditional methods. However, they still show limited performance by failing to model complex relations of side effects among drugs. Results To resolve the above problems, we propose a novel model by further incorporating complex relations of side effects into knowledge graph embeddings. Our model can translate and transmit multidirectional semantics with fewer parameters, leading to better scalability in large-scale knowledge graphs. Experimental evaluation shows that our model outperforms state-of-the-art models in terms of the average area under the ROC and precision-recall curves. Availability Code and data are available at: https://github.com/galaxysunwen/MSTE-master
Article
Artificial intelligence is an emerging sector in almost all fields. It is not confined only to a particular category and can be used in various fields like research, technology, and health. AI mainly concentrates on how computers analyze data and mimic the human thought process. As drug development involves high R & D costs and uncertainty in time consumption, artificial intelligence can serve as one of the promising solutions to overcome all these demerits. Due to the availability of enormous data, there are chances of missing out on some crucial details. For solving these issues, algorithms like machine learning, deep learning, and other expert systems are being used. On successful implementation of AI in the pharmaceutical field, the delays in drug development, and failure at the clinical and marketing level can be reduced. This review comprises information regarding the development of AI, its subfields, its overall implementation, and its application in the pharmaceutical sector and provides insights on challenges and limitations concerning AI.
Chapter
Drug discovery has predominantly followed the concept of “one drug, one target, one disease” for an extended period, case in point, to design chemical entities that can specifically bind to one key target related to a specific disease [1]. Collateral pharmacology aims to develop drugs that earmark multiple proteins or networks connected to diseases. It also demonstrates the possibility of finding multi-target and multi-component drugs that earmark disease-related networks at the system level [2]. Network pharmacology research integrates the data of various public databases, high-throughput screening (HTS), genome-wide association studies (GWAS), and large-scale omics (such as genomics, transcriptomics, metabonomics, and proteomics) to construct a network prediction or inference model. The analysis of the complex biological pathways influenced by drug therapy at different biological levels (molecules, cells, tissues, organs, and phenotypes) has given a boost to cognition of the biological mechanisms of complex diseases, the systemic mechanism of the impact of drugs, and the development of multi-target, multi-component drugs. This chapter selects some exceptional results of network pharmacology in the R&D and application of modern drugs in recent years, and analyzes the results from the dimensions of research purpose, data source, analysis index and algorithm, analysis results, experimental verification, and main conclusion, as a means to introduce the principal research contents, ideas, and procedures of frontier research in network pharmacology, for readers.
Article
Full-text available
Referring to common skeletal-related diseases, osteoporosis and Paget’s are two of the most frequently found diseases in the elderly. Nowadays, the combination of multiple drugs is the optimal therapy to decelerate osteoporosis and Paget’s pathologic process, which contains various underlying adverse effects due to drug-drug interactions (DDIs). Artificial intelligence (AI) has the potential to evaluate the interaction, pharmacodynamics, and possible side effects between drugs. In this research, we created an AI-based machine-learning model to predict the outcomes of interactions between drugs used for osteoporosis and Paget’s treatment, furthermore, to mitigate cost and time in implementing the best combination of medications in clinical practice. Our dataset was collected from the DrugBank database, and we then extracted a variety of chemical features from the simplified molecular-input line-entry system (SMILES) of defined drug pairs that interact with each other. Finally, machine-learning algorithms have been implemented to learn the extracted features. Our stack ensemble model from Random Forest and XGBoost reached an average accuracy of 74% in predicting DDIs. It was superior to individual models and previous methods in most measurement metrics. This study showed the potential of AI models in predicting DDIs of Osteoporosis-Paget’s disease in particular, and other diseases in general.
Article
COVID-19 is a worldwide health crisis seriously endangering the arsenal of antiviral and antibiotic drugs. It is urgent to find an effective antiviral drug against pandemic caused by the severe acute respiratory syndrome (Sars-Cov-2), which increases global health concerns. As it can be expensive and time-consuming to develop specific antiviral drugs, reuse of FDA-approved drugs that provide an opportunity to rapidly distribute effective therapeutics can allow to provide treatments with known preclinical, pharmacokinetic, pharmacodynamic and toxicity profiles that can quickly enter in clinical trials. In this study, using the structural information of molecules and proteins, a list of repurposed drug candidates was prepared again with the graph neural network-based GEFA model. The data set from the public databases DrugBank and PubChem were used for analysis. Using the Tanimoto/jaccard similarity analysis, a list of similar drugs was prepared by comparing the drugs used in the treatment of COVID-19 with the drugs used in the treatment of other diseases. The resultant drugs were compared with the drugs used in lung cancer and repurposed drugs were obtained again by calculating the binding strength between a drug and a target. The kinase inhibitors (erlotinib, lapatinib, vandetanib, pazopanib, cediranib, dasatinib, linifanib and tozasertib) obtained from the study can be used as an alternative for the treatment of COVID-19, as a combination of blocking agents (gefitinib, osimertinib, fedratinib, baricitinib, imatinib, sunitinib and ponatinib) such as ABL2, ABL1, EGFR, AAK1, FLT3 and JAK1, or antiviral therapies (ribavirin, ritonavir-lopinavir and remdesivir). Communicated by Ramaswamy H. Sarma
Article
Designing optimal combinatorial drug therapies is challenging, because the drug interactions depend not only on the drugs involved, but also on their doses. With recent advances, combinatorial drug therapy is closer than ever to clinical application. Herein, we summarize approaches and advances over the past decade for identifying and optimizing drug combination therapies, with innovations across research fields, covering physical laboratory platforms for combination screening to computational models and algorithms designed for synergism prediction and optimization. By comparing different types of approach, we detail a three-step workflow that could maximize the overall optimization efficiency, thus enabling the application of personalized optimization of combinatorial drug therapy.
Article
Information extraction (IE) is the process of automatically identifying structured information from unstructured or partially structured text. IE processes can involve several activities, such as named entity recognition, event extraction, relationship discovery, and document classification, with the overall goal of translating text into a more structured form. Information on the changes in the effect of a drug, when taken in combination with a second drug, is known as drug–drug interaction (DDI). DDIs can delay, decrease, or enhance absorption of drugs and thus decrease or increase their efficacy or cause adverse effects. Recent research trends have shown several adaptation of recurrent neural networks (RNNs) from text. In this study, we highlight significant challenges of using RNNs in biomedical text processing and propose automatic extraction of DDIs aiming at overcoming some challenges. Our results show that the system is competitive against other systems for the task of extracting DDIs.
Article
The advent of large-scale biomedical data and computational algorithms provides new opportunities for drug repurposing and discovery. It is of great interest to find an appropriate data representation and modeling method to facilitate these studies. The anatomical therapeutic chemical (ATC) classification system, proposed by the World Health Organization (WHO), is an essential source of information for drug repurposing and discovery. Besides, computational methods are applied to predict drug ATC classification. We conducted a systematic review of ATC computational prediction studies and revealed the differences in data sets, data representation, algorithm approaches, and evaluation metrics. We then proposed a deep fusion learning (DFL) framework to optimize the ATC prediction model, namely DeepATC. The methods based on graph convolutional network, inferring biological network and multimodel attentive fusion network were applied in DeepATC to extract the molecular topological information and low-dimensional representation from the molecular graph and heterogeneous biological networks. The results indicated that DeepATC achieved superior model performance with area under the curve (AUC) value at 0.968. Furthermore, the DFL framework was performed for the transcriptome data–based ATC prediction, as well as another independent task that is significantly relevant to drug discovery, namely drug–target interaction. The DFL-based model achieved excellent performance in the above-extended validation task, suggesting that the idea of aggregating the heterogeneous biological network and node’s (molecule or protein) self-topological features will bring inspiration for broader drug repurposing and discovery research.
Article
Drug–drug interaction (DDI) is a major public health problem contributing to 30% of the unexpected clinical adverse drug events. Informatics-based studies for DDI signal detection have been evolving in the last decade. We aim at providing a boosted machine learning (ML) framework to predict novel DDI safety signals with high precision. We propose a similarity-based machine learning framework called “SMDIP” using DrugBank as one of the most reliable pharmaceutical knowledge bases. For this study, DrugBank provides the latest drug information in terms of DDIs, targets, enzymes, transporters, and carriers. We computed drug–drug similarities using a Russell–Rao measure for the available biological and structural information on DrugBank for representing the sparse feature space. Logistic regression is adopted to conduct DDI classification with a focus on searching for key similarity predictors. Six types of ML models are deployed on the selected DDI key features. Our study reveals that SMDIP has yielded favourable predictive performance compared to relevant studies with results as follows: AUC 76%, precision 82%, accuracy 79%, recall 62%, specificity 90%, and F-measure 78%. To further confirm the reliability and reproducibility of SMDIP, we investigate SMDIP on an unseen subset of direct-acting-antiviral (DAA) drugs for treating hepatitis C infections. Forty novel DAA DDIs are predicted that show consistency with the pharmacokinetic and pharmacodynamic profiles of these drugs. Furthermore, several reports from the pharmacovigilance literature corroborate our framework results. Those evaluations show that SMDIP is a promising framework for uncovering DDIs, which can be multifariously feasible in drug development, postmarketing surveillance, and public health fields.
Article
Drug combination is a sensible strategy for disease treatment because it improves the treatment efficacy and reduces concomitant side effects. Due to the large number of possible combinations among candidate compounds, exhaustive screening is prohibitive. Currently, a large number of studies have focused on predicting potential drug combinations. However, these methods are not entirely satisfactory in terms of performance and scalability. In this paper, we proposed a Network Embedding frameWork in MultIplex Network (NEWMIN) to predict synthetic drug combinations. Based on a multiplex drug similarity network, we offered alternative methods to integrate useful information from different aspects and to decide the quantitative importance of each network. For drug combination prediction, we found seven novel drug combinations that have been validated by external sources among the top-ranked predictions of our model. To verify the feasibility of NEWMIN, we compared NEWMIN with other five methods, for which it showed better performance than other methods in terms of the area under the precision-recall curve and receiver operating characteristic curve.
Article
Drug repurposing refers to discovery of new medical instructions for existing chemical drugs, which has great pharmaceutical significance. Recently, large-scale biological datasets are increasingly available, and many graph neural network (GNN) based methods for drug repurposing have been developed. These methods often deem drug repurposing as a link prediction problem, which mines features of biological data to identify drug-disease associations (i.e., drug-disease links). Due to heterogeneity of data, we need to deeply explore heterogeneous information of biological network for drug repurposing. In this paper, we propose a Relation-aware Heterogeneous Graph Transformer (RHGT) model to capture heterogeneous information for drug repurposing. We first construct a drug-gene-disease interactive network-based on biological data, and then propose a three-level network embedding model, which learns network embeddings at fine-grained subtype-level, node-level and coarse-grained edge-level, respectively. The output of subtype-level is the input of node-level and edge-level, and the output of node-level is the input of edge level. We get edge embeddings at edge-level, which integrates edge type embeddings and node embeddings. We deem that in this way, characteristics of drug-gene-disease interactive network can be captured more comprehensively. Finally, we identify drug-disease associations (i.e., drug-disease links) based on the relationship between drug-gene edge embeddings and gene-disease edge embeddings. Experimental results show that our model performs better than other state-of-the-art graph neural network methods, which validates effectiveness of the proposed model.
Article
Full-text available
One of the main problems with the joint use of multiple drugs is that it may cause adverse drug interactions and side effects that damage the body. Therefore, it is important to predict potential drug interactions. However, most of the available prediction methods can only predict whether two drugs interact or not, whereas few methods can predict interaction events between two drugs. Accurately predicting interaction events of two drugs is more useful for researchers to study the mechanism of the interaction of two drugs. In the present study, we propose a novel method, MDF-SA-DDI, which predicts Downloaded from https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbab421/6406700 by guest on 21 October 2021 2 Lin et al. drug-drug interaction (DDI) events based on multi-source drug fusion, multi-source feature fusion and transformer self-attention mechanism. MDF-SA-DDI is mainly composed of two parts: multi-source drug fusion and multi-source feature fusion. First, we combine two drugs in four different ways and input the combined drug feature representation into four different drug fusion networks (Siamese network, convolutional neural network and two auto-encoders) to obtain the latent feature vectors of the drug pairs, in which the two auto-encoders have the same structure, and their main difference is the number of neurons in the input layer of the two auto-encoders. Then, we use transformer blocks that include self-attention mechanism to perform latent feature fusion. We conducted experiments on three different tasks with two datasets. On the small dataset, the area under the precision-recall-curve (AUPR) and F1 scores of our method on task 1 reached 0.9737 and 0.8878, respectively, which were better than the state-of-the-art method. On the large dataset, the AUPR and F1 scores of our method on task 1 reached 0.9773 and 0.9117, respectively. In task 2 and task 3 of two datasets, our method also achieved the same or better performance as the state-of-the-art method. More importantly, the case studies on five DDI events are conducted and achieved satisfactory performance. The source codes and data are available at https://gi thub.com/ShenggengLin/MDF-SA-DDI.
Article
Full-text available
ChEMBL is an open large-scale bioactivity database (https://www.ebi.ac.uk/chembl), previously described in the 2012 Nucleic Acids Research Database Issue. Since then, a variety of new data sources and improvements in functionality have contributed to the growth and utility of the resource. In particular, more comprehensive tracking of compounds from research stages through clinical development to market is provided through the inclusion of data from United States Adopted Name applications; a new richer data model for representing drug targets has been developed; and a number of methods have been put in place to allow users to more easily identify reliable data. Finally, access to ChEMBL is now available via a new Resource Description Framework format, in addition to the web-based interface, data downloads and web services.
Article
Full-text available
Sunitinib malate is a multitargeted receptor tyrosine kinase inhibitor used in the treatment of human malignancies. A substantial number of sunitinib-treated patients develop cardiac dysfunction, but the mechanism of sunitinib-induced cardiotoxicity is poorly understood. We show that mice treated with sunitinib develop cardiac and coronary microvascular dysfunction and exhibit an impaired cardiac response to stress. The physiological changes caused by treatment with sunitinib are accompanied by a substantial depletion of coronary microvascular pericytes. Pericytes are a cell type that is dependent on intact platelet-derived growth factor receptor (PDGFR) signaling but whose role in the heart is poorly defined. Sunitinib-induced pericyte depletion and coronary microvascular dysfunction are recapitulated by CP-673451, a structurally distinct PDGFR inhibitor, confirming the role of PDGFR in pericyte survival. Thalidomide, an anticancer agent that is known to exert beneficial effects on pericyte survival and function, prevents sunitinib-induced pericyte cell death in vitro and prevents sunitinib-induced cardiotoxicity in vivo in a mouse model. Our findings suggest that pericytes are the primary cellular target of sunitinib-induced cardiotoxicity and reveal the pericyte as a cell type of concern in the regulation of coronary microvascular function. Furthermore, our data provide preliminary evidence that thalidomide may prevent cardiotoxicity in sunitinib-treated cancer patients.
Article
Full-text available
There are numerous small molecular compounds around us to affect our health, such as drugs, pesticides, food additives, industrial chemicals, and environmental pollutants. Over decades, properties related to absorption, distribution, metabolism, excretion, and toxicity (ADMET) have become one of the most important issues to assess the effects or risks of these compounds on human body. Recent high-rate drug withdrawals increase the pressure on regulators and pharmaceutical industry to improve preclinical safety testing. Since in vivo and in vitro evaluations are costly and laborious, in silico techniques have been widely used to estimate these properties. In this review, we would briefly describe the recent advances of in silico ADMET prediction, with emphasis on substructure pattern recognition method that we developed recently. Challenges and limitions in the area of in silico ADMET prediction were further discussed, such as application domain of model, model validation techniques, global versus local models. At last, several new promising research directions were provided, such as computational systems toxicology (toxicogenomics), data-integration and meta-decision making systems, which could be used for systemic in silico ADMET prediction in drug discovery and hazard risk assessment.
Article
Full-text available
Drug-drug interactions (DDIs) can lead to serious and potentially lethal adverse events. In recent years, several drugs have been withdrawn from the market due to interaction-related adverse events (AEs). Current methods for detecting DDIs rely on the accumulation of sufficient clinical evidence in the post-market stage - a lengthy process that often takes years, during which time numerous patients may suffer from the adverse effects of the DDI. Detection methods are further hindered by the extremely large combinatoric space of possible drug-drug-AE combinations. There is therefore a practical need for predictive tools that can identify potential DDIs years in advance, enabling drug safety professionals to better prioritize their limited investigative resources and take appropriate regulatory action. To meet this need, we describe Predictive Pharmacointeraction Networks (PPINs) - a novel approach that predicts unknown DDIs by exploiting the network structure of all known DDIs, together with other intrinsic and taxonomic properties of drugs and AEs. We constructed an 856-drug DDI network from a 2009 snapshot of a widely-used drug safety database, and used it to develop PPIN models for predicting future DDIs. We compared the DDIs predicted based solely on these 2009 data, with newly reported DDIs that appeared in a 2012 snapshot of the same database. Using a standard multivariate approach to combine predictors, the PPIN model achieved an AUROC (area under the receiver operating characteristic curve) of 0.81 with a sensitivity of 48% given a specificity of 90%. An analysis of DDIs by severity level revealed that the model was most effective for predicting "contraindicated" DDIs (AUROC = 0.92) and less effective for "minor" DDIs (AUROC = 0.63). These results indicate that network based methods can be useful for predicting unknown drug-drug interactions.
Article
Full-text available
Identifying drug-drug interactions (DDIs) is a major challenge in drug development. Previous attempts have established formal approaches for pharmacokinetic (PK) DDIs, but there is not a feasible solution for pharmacodynamic (PD) DDIs because the endpoint is often a serious adverse event rather than a measurable change in drug concentration. Here, we developed a metric "S-score" that measures the strength of network connection between drug targets to predict PD DDIs. Utilizing known PD DDIs as golden standard positives (GSPs), we observed a significant correlation between S-score and the likelihood a PD DDI occurs. Our prediction was robust and surpassed existing methods as validated by two independent GSPs. Analysis of clinical side effect data suggested that the drugs having predicted DDIs have similar side effects. We further incorporated this clinical side effects evidence with S-score to increase the prediction specificity and sensitivity through a Bayesian probabilistic model. We have predicted 9,626 potential PD DDIs at the accuracy of 82% and the recall of 62%. Importantly, our algorithm provided opportunities for better understanding the potential molecular mechanisms or physiological effects underlying DDIs, as illustrated by the case studies.
Article
Full-text available
Despite considerable progress in genome- and proteome-based high-throughput screening methods and in rational drug design, the increase in approved drugs in the past decade did not match the increase of drug development costs. Network description and analysis not only gives a systems-level understanding of drug action and disease complexity, but can also help to improve the efficiency of drug design. We give a comprehensive assessment of the analytical tools of network topology and dynamics. The state-of-the-art use of chemical similarity, protein structure, protein-protein interaction, signaling, genetic interaction and metabolic networks in the discovery of drug targets is summarized. We propose that network targeting follows two basic strategies. The "central hit strategy" selectively targets central node/edges of the flexible networks of infectious agents or cancer cells to kill them. The "network influence strategy" works against other diseases, where an efficient reconfiguration of rigid networks needs to be achieved. It is shown how network techniques can help in the identification of single-target, edgetic, multi-target and allo-network drug target candidates. We review the recent boom in network methods helping hit identification, lead selection optimizing drug efficacy, as well as minimizing side-effects and drug toxicity. Successful network-based drug development strategies are shown through the examples of infections, cancer, metabolic diseases, neurodegenerative diseases and aging. Summarizing >1200 references we suggest an optimized protocol of network-aided drug development, and provide a list of systems-level hallmarks of drug quality. Finally, we highlight network-related drug development trends helping to achieve these hallmarks by a cohesive, global approach.
Article
Full-text available
Unexpected drug activities derived from off-targets are usually undesired and harmful; however, they can occasionally be beneficial for different therapeutic indications. There are many uncharacterized drugs whose target proteins (including the primary target and off-targets) remain unknown. The identification of all potential drug targets has become an important issue in drug repositioning to reuse known drugs for new therapeutic indications. We defined pharmacological similarity for all possible drugs using the US Food and Drug Administration's (FDA's) adverse event reporting system (AERS) and developed a new method to predict unknown drug-target interactions on a large scale from the integration of pharmacological similarity of drugs and genomic sequence similarity of target proteins in the framework of a pharmacogenomic approach. The proposed method was applicable to a large number of drugs and it was useful especially for predicting unknown drug-target interactions that could not be expected from drug chemical structures. We made a comprehensive prediction for potential off-targets of 1874 drugs with known targets and potential target profiles of 2519 drugs without known targets, which suggests many potential drug-target interactions that were not predicted by previous chemogenomic or pharmacogenomic approaches. Softwares are available upon request. yamanishi@bioreg.kyushu-u.ac.jp Datasets and all results are available at http://cbio.ensmp.fr/~yyamanishi/aers/.
Article
Full-text available
Drug-drug interactions (DDIs) are a common cause of adverse drug events. In this paper, we combined a literature discovery approach with analysis of a large electronic medical record database method to predict and evaluate novel DDIs. We predicted an initial set of 13197 potential DDIs based on substrates and inhibitors of cytochrome P450 (CYP) metabolism enzymes identified from published in vitro pharmacology experiments. Using a clinical repository of over 800,000 patients, we narrowed this theoretical set of DDIs to 3670 drug pairs actually taken by patients. Finally, we sought to identify novel combinations that synergistically increased the risk of myopathy. Five pairs were identified with their p-values less than 1E-06: loratadine and simvastatin (relative risk or RR = 1.69); loratadine and alprazolam (RR = 1.86); loratadine and duloxetine (RR = 1.94); loratadine and ropinirole (RR = 3.21); and promethazine and tegaserod (RR = 3.00). When taken together, each drug pair showed a significantly increased risk of myopathy when compared to the expected additive myopathy risk from taking either of the drugs alone. Based on additional literature data on in vitro drug metabolism and inhibition potency, loratadine and simvastatin and tegaserod and promethazine were predicted to have a strong DDI through the CYP3A4 and CYP2D6 enzymes, respectively. This new translational biomedical informatics approach supports not only detection of new clinically significant DDI signals, but also evaluation of their potential molecular mechanisms.
Article
Full-text available
Understanding drug bioactivities is crucial for early-stage drug discovery, toxicology studies and clinical trials. Network pharmacology is a promising approach to better understand the molecular mechanisms of drug bioactivities. With a dramatic increase of rich data sources that document drugs' structural, chemical, and biological activities, it is necessary to develop an automated tool to construct a drug-target network for candidate drugs, thus facilitating the drug discovery process. We designed a computational workflow to construct drug-target networks from different knowledge bases including DrugBank, PharmGKB, and the PINA database. To automatically implement the workflow, we created a web-based tool called DTome (Drug-Target interactome tool), which is comprised of a database schema and a user-friendly web interface. The DTome tool utilizes web-based queries to search candidate drugs and then construct a DTome network by extracting and integrating four types of interactions. The four types are adverse drug interactions, drug-target interactions, drug-gene associations, and target-/gene-protein interactions. Additionally, we provided a detailed network analysis and visualization process to illustrate how to analyze and interpret the DTome network. The DTome tool is publicly available at http://bioinfo.mc.vanderbilt.edu/DTome. As demonstrated with the antipsychotic drug clozapine, the DTome tool was effective and promising for the investigation of relationships among drugs, adverse interaction drugs, drug primary targets, drug-associated genes, and proteins directly interacting with targets or genes. The resultant DTome network provides researchers with direct insights into their interest drug(s), such as the molecular mechanisms of drug actions. We believe such a tool can facilitate identification of drug targets and drug adverse interactions.
Article
Full-text available
Chemical-protein interaction (CPI) is the central topic of target identification and drug discovery. However, large scale determination of CPI is a big challenge for in vitro or in vivo experiments, while in silico prediction shows great advantages due to low cost and high accuracy. On the basis of our previous drug-target interaction prediction via network-based inference (NBI) method, we further developed node- and edge-weighted NBI methods for CPI prediction here. Two comprehensive CPI bipartite networks extracted from ChEMBL database were used to evaluate the methods, one containing 17,111 CPI pairs between 4,741 compounds and 97 G protein-coupled receptors, the other including 13,648 CPI pairs between 2,827 compounds and 206 kinases. The range of the area under receiver operating characteristic curves was 0.73 to 0.83 for the external validation sets, which confirmed the reliability of the prediction. The weak-interaction hypothesis in CPI network was identified by the edge-weighted NBI method. Moreover, to validate the methods, several candidate targets were predicted for five approved drugs, namely imatinib, dasatinib, sertindole, olanzapine and ziprasidone. The molecular hypotheses and experimental evidence for these predictions were further provided. These results confirmed that our methods have potential values in understanding molecular basis of drug polypharmacology and would be helpful for drug repositioning.
Article
Full-text available
Inferring drug-drug interactions (DDIs) is an essential step in drug development and drug administration. Most computational inference methods focus on modeling drug pharmacokinetics, aiming at interactions that result from a common metabolizing enzyme (CYP). Here, we introduce a novel prediction method, INDI (INferring Drug Interactions), allowing the inference of both pharmacokinetic, CYP-related DDIs (along with their associated CYPs) and pharmacodynamic, non-CYP associated ones. On cross validation, it obtains high specificity and sensitivity levels (AUC (area under the receiver-operating characteristic curve) ≥0.93). In application to the FDA adverse event reporting system, 53% of the drug events could potentially be connected to known (41%) or predicted (12%) DDIs. Additionally, INDI predicts the severity level of each DDI upon co-administration of the involved drugs, suggesting that severe interactions are abundant in the clinical practice. Examining regularly taken medications by hospitalized patients, 18% of the patients receive known or predicted severely interacting drugs and are hospitalized more frequently. Access to INDI and its predictions is provided via a web tool at http://www.cs.tau.ac.il/~bnet/software/INDI, facilitating the inference and exploration of drug interactions and providing important leads for physicians and pharmaceutical companies alike.
Article
Full-text available
Objective Adverse drug reaction (ADR) is one of the major causes of failure in drug development. Severe ADRs that go undetected until the post-marketing phase of a drug often lead to patient morbidity. Accurate prediction of potential ADRs is required in the entire life cycle of a drug, including early stages of drug design, different phases of clinical trials, and post-marketing surveillance. Methods Many studies have utilized either chemical structures or molecular pathways of the drugs to predict ADRs. Here, the authors propose a machine-learning-based approach for ADR prediction by integrating the phenotypic characteristics of a drug, including indications and other known ADRs, with the drug's chemical structures and biological properties, including protein targets and pathway information. A large-scale study was conducted to predict 1385 known ADRs of 832 approved drugs, and five machine-learning algorithms for this task were compared. Results This evaluation, based on a fivefold cross-validation, showed that the support vector machine algorithm outperformed the others. Of the three types of information, phenotypic data were the most informative for ADR prediction. When biological and phenotypic features were added to the baseline chemical information, the ADR prediction model achieved significant improvements in area under the curve (from 0.9054 to 0.9524), precision (from 43.37% to 66.17%), and recall (from 49.25% to 63.06%). Most importantly, the proposed model successfully predicted the ADRs associated with withdrawal of rofecoxib and cerivastatin. Conclusion The results suggest that phenotypic information on drugs is valuable for ADR prediction. Moreover, they demonstrate that different models that combine chemical, biological, or phenotypic information can be built from approved drugs, and they have the potential to detect clinically important ADRs in both preclinical and post-marketing phases.
Article
Full-text available
Author Summary Study of drug-target interaction is an important topic toward elucidation of protein functions and understanding of molecular mechanisms inside cells. Traditional methods to predict new targets for known drugs were based on small molecules, protein targets or phenotype features. Here, we proposed a network-based inference (NBI) method which only used drug-target bipartite network topology similarity to infer new targets for known drugs. The performance of NBI outperformed the drug-based similarity inference and target-based similarity inference methods as well as other published methods. Via the NBI method five old drugs, namely montelukast, diclofenac, simvastatin, ketoconazole, and itraconazole, were identified to have polypharmacological effects on human estrogen receptors or dipeptidyl peptidase-IV with half maximal inhibitory or effective concentration from submicromolar to micromolar by in vitro assays. Moreover, simvastatin and ketoconazole showed potent antiproliferative activities on human MDA-MB-231 breast cancer cell line in MTT assays. The results indicated that the drug-target bipartite network-based inference method could be a useful tool for fishing novel drug-target interactions in molecular polypharmacological space.
Article
Full-text available
Adverse drug events remain a leading cause of morbidity and mortality around the world. Many adverse events are not detected during clinical trials before a drug receives approval for use in the clinic. Fortunately, as part of postmarketing surveillance, regulatory agencies and other institutions maintain large collections of adverse event reports, and these databases present an opportunity to study drug effects from patient population data. However, confounding factors such as concomitant medications, patient demographics, patient medical histories, and reasons for prescribing a drug often are uncharacterized in spontaneous reporting systems, and these omissions can limit the use of quantitative signal detection methods used in the analysis of such data. Here, we present an adaptive data-driven approach for correcting these factors in cases for which the covariates are unknown or unmeasured and combine this approach with existing methods to improve analyses of drug effects using three test data sets. We also present a comprehensive database of drug effects (Offsides) and a database of drug-drug interaction side effects (Twosides). To demonstrate the biological use of these new resources, we used them to identify drug targets, predict drug indications, and discover drug class interactions. We then corroborated 47 (P < 0.0001) of the drug class interactions using an independent analysis of electronic medical records. Our analysis suggests that combined treatment with selective serotonin reuptake inhibitors and thiazides is associated with significantly increased incidence of prolonged QT intervals. We conclude that confounding effects from covariates in observational clinical data can be controlled in data analyses and thus improve the detection and prediction of adverse drug effects and interactions.
Article
Full-text available
ABSTRACT: A frequent problem in computational modeling is the interconversion of chemical structures between different formats. While standard interchange formats exist (for example, Chemical Markup Language) and de facto standards have arisen (for example, SMILES format), the need to interconvert formats is a continuing problem due to the multitude of different application areas for chemistry data, differences in the data stored by different formats (0D versus 3D, for example), and competition between software along with a lack of vendor-neutral formats. We discuss, for the first time, Open Babel, an open-source chemical toolbox that speaks the many languages of chemical data. Open Babel version 2.3 interconverts over 110 formats. The need to represent such a wide variety of chemical and molecular data requires a library that implements a wide range of cheminformatics algorithms, from partial charge assignment and aromaticity detection, to bond order perception and canonicalization. We detail the implementation of Open Babel, describe key advances in the 2.3 release, and outline a variety of uses both in terms of software products and scientific research, including applications far beyond simple format interconversion. Open Babel presents a solution to the proliferation of multiple chemical file formats. In addition, it provides a variety of useful utilities from conformer searching and 2D depiction, to filtering, batch conversion, and substructure and similarity searching. For developers, it can be used as a programming library to handle chemical data in areas such as organic chemistry, drug design, materials science, and computational chemistry. It is freely available under an open-source license from http://openbabel.org.
Article
Full-text available
Knowledge and investigation of therapeutic targets (responsible for drug efficacy) and the targeted drugs facilitate target and drug discovery and validation. Therapeutic Target Database (TTD, http://bidd.nus.edu.sg/group/ttd/ttd.asp) has been developed to provide comprehensive information about efficacy targets and the corresponding approved, clinical trial and investigative drugs. Since its last update, major improvements and updates have been made to TTD. In addition to the significant increase of data content (from 1894 targets and 5028 drugs to 2025 targets and 17,816 drugs), we added target validation information (drug potency against target, effect against disease models and effect of target knockout, knockdown or genetic variations) for 932 targets, and 841 quantitative structure activity relationship models for active compounds of 228 chemical types against 121 targets. Moreover, we added the data from our previous drug studies including 3681 multi-target agents against 108 target pairs, 116 drug combinations with their synergistic, additive, antagonistic, potentiative or reductive mechanisms, 1427 natural product-derived approved, clinical trial and pre-clinical drugs and cross-links to the clinical trial information page in the ClinicalTrials.gov database for 770 clinical trial drugs. These updates are useful for facilitating target discovery and validation, drug lead discovery and optimization, and the development of multi-target drugs and drug combinations.
Article
Full-text available
Adverse drug events (ADEs) are common and account for 770 000 injuries and deaths each year and drug interactions account for as much as 30% of these ADEs. Spontaneous reporting systems routinely collect ADEs from patients on complex combinations of medications and provide an opportunity to discover unexpected drug interactions. Unfortunately, current algorithms for such "signal detection" are limited by underreporting of interactions that are not expected. We present a novel method to identify latent drug interaction signals in the case of underreporting. We identified eight clinically significant adverse events. We used the FDA's Adverse Event Reporting System to build profiles for these adverse events based on the side effects of drugs known to produce them. We then looked for pairs of drugs that match these single-drug profiles in order to predict potential interactions. We evaluated these interactions in two independent data sets and also through a retrospective analysis of the Stanford Hospital electronic medical records. We identified 171 novel drug interactions (for eight adverse event categories) that are significantly enriched for known drug interactions (p=0.0009) and used the electronic medical record for independently testing drug interaction hypotheses using multivariate statistical models with covariates. Our method provides an option for detecting hidden interactions in spontaneous reporting systems by using side effect profiles to infer the presence of unreported adverse events.
Article
Full-text available
DrugBank (http://www.drugbank.ca) is a richly annotated database of drug and drug target information. It contains extensive data on the nomenclature, ontology, chemistry, structure, function, action, pharmacology, pharmacokinetics, metabolism and pharmaceutical properties of both small molecule and large molecule (biotech) drugs. It also contains comprehensive information on the target diseases, proteins, genes and organisms on which these drugs act. First released in 2006, DrugBank has become widely used by pharmacists, medicinal chemists, pharmaceutical researchers, clinicians, educators and the general public. Since its last update in 2008, DrugBank has been greatly expanded through the addition of new drugs, new targets and the inclusion of more than 40 new data fields per drug entry (a 40% increase in data 'depth'). These data field additions include illustrated drug-action pathways, drug transporter data, drug metabolite data, pharmacogenomic data, adverse drug response data, ADMET data, pharmacokinetic data, computed property data and chemical classification data. DrugBank 3.0 also offers expanded database links, improved search tools for drug-drug and food-drug interaction, new resources for querying and viewing drug pathways and hundreds of new drug entries with detailed patent, pricing and manufacturer data. These additions have been complemented by enhancements to the quality and quantity of existing data, particularly with regard to drug target, drug description and drug action data. DrugBank 3.0 represents the result of 2 years of manual annotation work aimed at making the database much more useful for a wide range of 'omics' (i.e. pharmacogenomic, pharmacoproteomic, pharmacometabolomic and even pharmacoeconomic) applications.
Article
Full-text available
The Comparative Toxicogenomics Database (CTD) is a public resource that promotes understanding about the interaction of environmental chemicals with gene products, and their effects on human health. Biocurators at CTD manually curate a triad of chemical–gene, chemical–disease and gene–disease relationships from the literature. These core data are then integrated to construct chemical–gene–disease networks and to predict many novel relationships using different types of associated data. Since 2009, we dramatically increased the content of CTD to 1.4 million chemical–gene–disease data points and added many features, statistical analyses and analytical tools, including GeneComps and ChemComps (to find comparable genes and chemicals that share toxicogenomic profiles), enriched Gene Ontology terms associated with chemicals, statistically ranked chemical–disease inferences, Venn diagram tools to discover overlapping and unique attributes of any set of chemicals, genes or disease, and enhanced gene pathway data content, among other features. Together, this wealth of expanded chemical–gene–disease data continues to help users generate testable hypotheses about the molecular mechanisms of environmental diseases. CTD is freely available at http://ctd.mdibl.org.
Article
Full-text available
Motivation: Identifying drug–drug interactions (DDIs) is a critical process in drug administration and drug development. Clinical support tools often provide comprehensive lists of DDIs, but they usually lack the supporting scientific evidences and different tools can return inconsistent results. In this article, we propose a novel approach that integrates text mining and automated reasoning to derive DDIs. Through the extraction of various facts of drug metabolism, not only the DDIs that are explicitly mentioned in text can be extracted but also the potential interactions that can be inferred by reasoning. Results: Our approach was able to find several potential DDIs that are not present in DrugBank. We manually evaluated these interactions based on their supporting evidences, and our analysis revealed that 81.3% of these interactions are determined to be correct. This suggests that our approach can uncover potential DDIs with scientific evidences explaining the mechanism of the interactions. Contact: luis.tari@roche.com
Article
Full-text available
The molecular understanding of phenotypes caused by drugs in humans is essential for elucidating mechanisms of action and for developing personalized medicines. Side effects of drugs (also known as adverse drug reactions) are an important source of human phenotypic information, but so far research on this topic has been hampered by insufficient accessibility of data. Consequently, we have developed a public, computer-readable side effect resource (SIDER) that connects 888 drugs to 1450 side effect terms. It contains information on frequency in patients for one-third of the drug-side effect pairs. For 199 drugs, the side effect frequency of placebo administration could also be extracted. We illustrate the potential of SIDER with a number of analyses. The resource is freely available for academic research at http://sideeffects.embl.de.
Article
Full-text available
Much of the information on the Cytochrome P450 enzymes (CYPs) is spread across literature and the internet. Aggregating knowledge about CYPs into one database makes the search more efficient. Text mining on 57 CYPs and drugs led to a mass of papers, which were screened manually for facts about metabolism, SNPs and their effects on drug degradation. Information was put into a database, which enables the user not only to look up a particular CYP and all metabolized drugs, but also to check tolerability of drug-cocktails and to find alternative combinations, to use metabolic pathways more efficiently. The SuperCYP database contains 1170 drugs with more than 3800 interactions including references. Approximately 2000 SNPs and mutations are listed and ordered according to their effect on expression and/or activity. SuperCYP (http://bioinformatics.charite.de/supercyp) is a comprehensive resource focused on CYPs and drug metabolism. Homology-modeled structures of the CYPs can be downloaded in PDB format and related drugs are available as MOL-files. Within the resource, CYPs can be aligned with each other, drug-cocktails can be ‘mixed’, SNPs, protein point mutations, and their effects can be viewed and corresponding PubMed IDs are given. SuperCYP is meant to be a platform and a starting point for scientists and health professionals for furthering their research.
Article
Full-text available
The cytochrome P450 (CYP) gene family catalyzes drug metabolism and bioactivation and is therefore relevant to drug development. We determined potency values for 17,143 compounds against five recombinant CYP isozymes (1A2, 2C9, 2C19, 2D6 and 3A4) using an in vitro bioluminescent assay. The compounds included libraries of US Food and Drug Administration (FDA)-approved drugs and screening libraries. We observed cross-library isozyme inhibition (30-78%) with important differences between libraries. Whereas only 7% of the typical screening library was inactive against all five isozymes, 33% of FDA-approved drugs were inactive, reflecting the optimized pharmacological properties of the latter. Our results suggest that low CYP 2C isozyme activity is a common property of drugs, whereas other isozymes, such as CYP 2D6, show little discrimination between drugs and unoptimized compounds found in screening libraries. We also identified chemical substructures that differentiated between the five isozymes. The pharmacological compendium described here should further the understanding of CYP isozymes.
Article
Full-text available
To identify and evaluate the systems failures that underlie errors causing adverse drug events (ADEs) and potential ADEs. Systems analysis of events from a prospective cohort study. All admissions to 11 medical and surgical units in two tertiary care hospitals over a 6-month period. Errors, proximal causes, and systems failures. Errors were detected by interviews of those involved. Errors were classified according to proximal cause and underlying systems failure by multidisciplinary teams of physicians, nurses, pharmacists, and systems analysts. During this period, 334 errors were detected as the causes of 264 preventable ADEs and potential ADEs. Sixteen major systems failures were identified as the underlying causes of the errors. The most common systems failure was in the dissemination of drug knowledge, particularly to physicians, accounting for 29% of the 334 errors. Inadequate availability of patient information, such as the results of laboratory tests, was associated with 18% of errors. Seven systems failures accounted for 78% of the errors; all could be improved by better information systems. Hospital personnel willingly participated in the detection and investigation of drug use errors and were able to identify underlying systems failures. The most common defects were in systems to disseminate knowledge about drugs and to make drug and patient information readily accessible at the time it is needed. Systems changes to improve dissemination and display of drug and patient data should make errors in the use of drugs less likely.
Article
Full-text available
The purpose of this study was to estimate and compare the effects of antipsychotics-both the newer ones and the conventional ones-on body weight. Method: A comprehensive literature search identified 81 English- and non-English-language articles that included data on weight change in antipsychotic-treated patients. For each agent, a meta-analysis and random effects metaregression estimated the weight change after 10 weeks of treatment at a standard dose. A comprehensive narrative review was also conducted on all articles that did not yield quantitative information but did yield important qualitative information. Placebo was associated with a mean weight reduction of 0.74 kg. Among conventional agents, mean weight change ranged from a reduction of 0.39 kg with molindone to an increase of 3.19 kg with thioridazine. Among newer antipsychotic agents, mean increases were as follows: clozapine, 4.45 kg; olanzapine, 4.15 kg; sertindole, 2.92 kg; risperidone, 2.10 kg; and ziprasidone, 0.04 kg. Insufficient data were available to evaluate quetiapine at 10 weeks. Both conventional and newer antipsychotics are associated with weight gain. Among the newer agents, clozapine appears to have the greatest potential to induce weight gain, and ziprasidone the least. The differences among newer agents may affect compliance with medication and health risk.
Article
Full-text available
We provide a unified overview of methods that currently are widely used to assess the accuracy of prediction algorithms, from raw percentages, quadratic error measures and other distances, and correlation coefficients, and to information theoretic measures such as relative entropy and mutual information. We briefly discuss the advantages and disadvantages of each approach. For classification tasks, we derive new learning algorithms for the design of prediction systems by directly optimising the correlation coefficient. We observe and prove several results relating sensitivity and specificity of optimal systems. While the principles are general, we illustrate the applicability on specific problems such as protein secondary structure and signal peptide prediction.
Article
Full-text available
Cytoscape is an open source software project for integrating biomolecular interaction networks with high-throughput expression data and other molecular states into a unified conceptual framework. Although applicable to any system of molecular components and interactions, Cytoscape is most powerful when used in conjunction with large databases of protein-protein, protein-DNA, and genetic interactions that are increasingly available for humans and model organisms. Cytoscape's software Core provides basic functionality to layout and query the network; to visually integrate the network with expression profiles, phenotypes, and other molecular states; and to link the network to databases of functional annotations. The Core is extensible through a straightforward plug-in architecture, allowing rapid development of additional computational analyses and features. Several case studies of Cytoscape plug-ins are surveyed, including a search for interaction pathways correlating with changes in gene expression, a study of protein complexes involved in cellular recovery to DNA damage, inference of a combined physical/functional interaction network for Halobacterium, and an interface to detailed stochastic/kinetic gene regulatory models.
Article
Full-text available
Olanzapine, an atypical antipsychotic, is often regarded as a safe choice for psychosis management. We hereby report an aged case that presented with conscious depression, bradycardia, hypotension, miosis and hypothermia. Olanzapine was thought to be the offending agent. His condition improved with supportive therapy.
Article
Full-text available
In vitro screening for drugs that inhibit cytochrome P450 enzymes is well established as a means for predicting potential metabolism-mediated drug interactions in vivo. Given that these predictions are based on enzyme kinetic parameters observed from in vitro experiments, the miscalculation of the inhibitory potency of a compound can lead to an inaccurate prediction of an in vivo drug interaction, potentially precluding a safe drug from advancing in development or allowing a potent inhibitor to 'slip' into the patient population. Here, we describe the principles underlying the generation of in vitro drug metabolism data and highlight commonly encountered uncertainties and sources of bias and error that can affect extrapolation of drug-drug interaction information to the clinical setting.
Article
Thesupport-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
Article
Objective —To identify and evaluate the systems failures that underlie errors causing adverse drug events (ADEs) and potential ADEs. Design —Systems analysis of events from a prospective cohort study. Participants —All admissions to 11 medical and surgical units in two tertiary care hospitals over a 6-month period. Main Outcome Measures —Errors, proximal causes, and systems failures. Methods —Errors were detected by interviews of those involved. Errors were classified according to proximal cause and underlying systems failure by multidisciplinary teams of physicians, nurses, pharmacists, and systems analysts. Results —During this period, 334 errors were detected as the causes of 264 preventable ADEs and potential ADEs. Sixteen major systems failures were identified as the underlying causes of the errors. The most common systems failure was in the dissemination of drug knowledge, particularly to physicians, accounting for 29% of the 334 errors. Inadequate availability of patient information, such as the results of laboratory tests, was associated with 18% of errors. Seven systems failures accounted for 78% of the errors; all could be improved by better information systems. Conclusions —Hospital personnel willingly participated in the detection and investigation of drug use errors and were able to identify underlying systems failures. The most common defects were in systems to disseminate knowledge about drugs and to make drug and patient information readily accessible at the time it is needed. Systems changes to improve dissemination and display of drug and patient data should make errors in the use of drugs less likely.(JAMA. 1995;274:35-43)