A Review of Approaches for Predicting Drug–Drug Interactions Based on Machine Learning

Han, Koeun; Cao, Peigang; Wang, Yu; Xie, Fang; Ma, Jiaqi; Yu, Mengyao; Wang, Jianchun; Xu, Yaoqun; Zhang, Yunlong; Wan, Jie

doi:10.3389/fphar.2021.814858

Cited by 72 publications

(44 citation statements)

References 87 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Thus, additional studies are warranted to examine how pDDI resources can be better integrated in routine clinical practice to provide a quick overview on unwanted effects and serious problems related to inappropriate drug use in MS patients. In the future, patient safety might be improved by machine learning methods, which can help in predicting relevant interactions between multiple drugs (Basile et al, 2019;Han et al, 2022). Further research might also involve the patients and investigate whether they are aware of the problem and understand information about pDDIs (Hammar et al, 2021).…”

Section: Discussionmentioning

confidence: 99%

Screening for severe drug-drug interactions in patients with multiple sclerosis: A comparison of three drug interaction databases

et al. 2022

View full text Add to dashboard Cite

Background: Patients with multiple sclerosis (MS) often undergo complex treatment regimens, resulting in an increased risk of polypharmacy and potential drug-drug interactions (pDDIs). Drug interaction databases are useful for identifying pDDIs to support safer medication use.Objective: To compare three different screening tools regarding the detection and classification of pDDIs in a cohort of MS patients. Furthermore, we aimed at ascertaining sociodemographic and clinical factors that are associated with the occurrence of severe pDDIs.Methods: The databases Stockley’s, Drugs.com and MediQ were used to identify pDDIs by screening the medication schedules of 627 patients. We determined the overlap of the identified pDDIs and the level of agreement in pDDI severity ratings between the three databases. Logistic regression analyses were conducted to determine patient risk factors of having a severe pDDI.Results: The most different pDDIs were identified using MediQ (n = 1,161), followed by Drugs.com (n = 923) and Stockley’s (n = 706). The proportion of pDDIs classified as severe was much higher for Stockley’s (37.4%) than for Drugs.com (14.4%) and MediQ (0.9%). Overall, 1,684 different pDDIs were identified by at least one database, of which 318 pDDIs (18.9%) were detected with all three databases. Only 55 pDDIs (3.3%) have been reported with the same severity level across all databases. A total of 336 pDDIs were classified as severe (271 pDDIs by one database, 59 by two databases and 6 by three databases). Stockley’s and Drugs.com revealed 47 and 23 severe pDDIs, respectively, that were not included in the other databases. At least one severe pDDI was found for 35.2% of the patients. The most common severe pDDI was the combination of acetylsalicylic acid with enoxaparin, and citalopram was the drug most frequently involved in different severe pDDIs. The strongest predictors of having a severe pDDI were a greater number of drugs taken, an older age, living alone, a higher number of comorbidities and a lower educational level.Conclusions: The information on pDDIs are heterogeneous between the databases examined. More than one resource should be used in clinical practice to evaluate pDDIs. Regular medication reviews and exchange of information between treating physicians can help avoid severe pDDIs.

show abstract

Section: Discussionmentioning

confidence: 99%

Screening for severe drug-drug interactions in patients with multiple sclerosis: A comparison of three drug interaction databases

et al. 2022

View full text Add to dashboard Cite

show abstract

“…A desirable DDI prediction model should not only pursue good prediction accuracy but also pursue the ability to accurately predict the types of drug interactions [ 45 ]. Therefore, this experiment conducts 65 types of DDI event prediction among 572 drugs and 37,264 drug pairs with known interactions.…”

Section: Resultsmentioning

confidence: 99%

Multitype drug interaction prediction based on the deep fusion of drug features and topological relationships

et al. 2022

View full text Add to dashboard Cite

Drug–drug interaction (DDI) prediction has received considerable attention from industry and academia. Most existing methods predict DDIs from drug attributes or relationships with neighbors, which does not guarantee that informative drug embeddings for prediction will be obtained. To address this limitation, we propose a multitype drug interaction prediction method based on the deep fusion of drug features and topological relationships, abbreviated DM-DDI. The proposed method adopts a deep fusion strategy to combine drug features and topologies to learn representative drug embeddings for DDI prediction. Specifically, a deep neural network model is first used on the drug feature matrix to extract feature information, while a graph convolutional network model is employed to capture structural information from the adjacency matrix. Then, we adopt delivery operations that allow the two models to exchange information between layers, as well as an attention mechanism for a weighted fusion of the two learned embeddings before the output layer. Finally, the unified drug embeddings for the downstream task are obtained. We conducted extensive experiments on real-world datasets, the experimental results demonstrated that DM-DDI achieved more accurate prediction results than state-of-the-art baselines. Furthermore, in two tasks that are more similar to real-world scenarios, DM-DDI outperformed other prediction methods for unknown drugs.

show abstract

“…Most of the alerts generated by the legacy CDSS were related to DDIs and dosages [ 26 ]. Although there are theoretical and review ML studies on DDI extraction from the biomedical literature [ 27 ], DrugBank and other databases [ 28 , 29 ], bioinformatics algorithms to predict DDI [ 30 ], and clinical safety DDI information retrieval [ 31 ], there are no real-life studies that reflect clinical practice in neonates.…”

Section: Discussionmentioning

confidence: 99%

Novel Method for Early Prediction of Clinically Significant Drug–Drug Interactions with a Machine Learning Algorithm Based on Risk Matrix Analysis in the NICU

Yalçın

Kasikci

Çelik

et al. 2022

JCM

View full text Add to dashboard Cite

Aims: Evidence for drug–drug interactions (DDIs) that may cause age-dependent differences in the incidence and severity of adverse drug reactions (ADRs) in newborns is sparse. We aimed to develop machine learning (ML) algorithms that predict DDI presence by integrating each DDI, which is objectively evaluated with the scales in a risk matrix (probability + severity). Methods: This double-center, prospective randomized cohort study included neonates admitted to the neonatal intensive care unit in a tertiary referral hospital during the 17-month study period. Drugs were classified by the Anatomical Therapeutic Chemical (ATC) classification and assessed for potential and clinically relevant DDIs to risk analyses with the Drug Interaction Probability Scale (DIPS, causal probability) and the Lexicomp® DDI (severity) database. Results: A total of 412 neonates (median (interquartile range) gestational age of 37 (4) weeks) were included with 32,925 patient days, 131 different medications, and 11,908 medication orders. Overall, at least one potential DDI was observed in 125 (30.4%) of the patients (2.6 potential DDI/patient). A total of 38 of these 125 patients had clinically relevant DDIs causing adverse drug reactions (2.0 clinical DDI/patient). The vast majority of these DDIs (90.66%) were assessed to be at moderate risk. The performance of the ML algorithms that predicts of the presence of relevant DDI was as follows: accuracy 0.944 (95% CI 0.888–0.972), sensitivity 0.892 (95% CI 0.769–0.962), F1 score 0.904, and AUC 0.929 (95% CI 0.874–0.983). Conclusions: In clinical practice, it is expected that optimization in treatment can be achieved with the implementation of this high-performance web tool, created to predict DDIs before they occur with a newborn-centered approach.

show abstract

A Review of Approaches for Predicting Drug–Drug Interactions Based on Machine Learning

Cited by 72 publications

References 87 publications

Screening for severe drug-drug interactions in patients with multiple sclerosis: A comparison of three drug interaction databases

Screening for severe drug-drug interactions in patients with multiple sclerosis: A comparison of three drug interaction databases

Multitype drug interaction prediction based on the deep fusion of drug features and topological relationships

Novel Method for Early Prediction of Clinically Significant Drug–Drug Interactions with a Machine Learning Algorithm Based on Risk Matrix Analysis in the NICU

Contact Info

Product

Resources

About