Prediction Of Adverse Drug Reactions Research Articles

[Artículo traducido] Predicción de reacciones adversas a medicamentos en pacientes geriátricos ingresados en unidades de cuidados intensivos

IntroductionIntensive Care Units (ICUs) pose challenges in managing critically-ill patients with polypharmacy, potentially leading to Adverse Drug Reactions (ADRs), particularly in the elderly. ObjectiveTo evaluate whether the severity and clinical prognosis scores used in ICUs correlate with the prediction of ADRs in aged patients admitted to an ICU. MethodsA cohort study was conducted in a Brazilian University Hospital ICU. APACHE II and SAPS 3 assessed clinical prognosis, while GerontoNet ADR Risk Score and BADRI evaluated ADR risk at ICU admission. Severity of the patients' clinical conditions was evaluated daily based on the SOFA score. Adverse Drug Reaction (ADR) screening was performed daily through the identification of ADR triggers. Results1295 triggers were identified (median 30 per patient, IQR = 28), with 15 suspected ADRs. No correlation was observed between patient severity and ADRs at admission (p=0.26), during hospitalization (p=0.91), or at follow-up (p=0.77). There was also no association between death and ADRs (p=0.28) or worse prognosis and ADRs (p>0.05). Higher BADRI scores correlated with more ADRs (p=0.001). ConclusionsThe data suggest that employing the severity and clinical prognosis scores used in Intensive Care Units is not sufficient to direct active pharmacovigilance efforts, which are therefore indicated for critically ill patients.

Advancing Adverse Drug Reaction Prediction with Deep Chemical Language Model for Drug Safety Evaluation.

The accurate prediction of adverse drug reactions (ADRs) is essential for comprehensive drug safety evaluation. Pre-trained deep chemical language models have emerged as powerful tools capable of automatically learning molecular structural features from large-scale datasets, showing promising capabilities for the downstream prediction of molecular properties. However, the performance of pre-trained chemical language models in predicting ADRs, especially idiosyncratic ADRs induced by marketed drugs, remains largely unexplored. In this study, we propose MoLFormer-XL, a pre-trained model for encoding molecular features from canonical SMILES, in conjunction with a CNN-based model to predict drug-induced QT interval prolongation (DIQT), drug-induced teratogenicity (DIT), and drug-induced rhabdomyolysis (DIR). Our results demonstrate that the proposed model outperforms conventional models applied in previous studies for predicting DIQT, DIT, and DIR. Notably, an analysis of the learned linear attention maps highlights amines, alcohol, ethers, and aromatic halogen compounds as strongly associated with the three types of ADRs. These findings hold promise for enhancing drug discovery pipelines and reducing the drug attrition rate due to safety concerns.

NEURAL NETWORK-BASED ADVERSE DRUG REACTION PREDICTION USING MOLECULAR SUBSTRUCTURE ANALYSES

Objective: This study aims to enhance early detection and prediction by exploiting drug molecular substructures, overcoming challenges posed by limited authentic patient data in the medical domain. Methods: The study implemented a neural network approach to optimize molecular fingerprint algorithms and employed various machine learning algorithms for predictions. Additionally, the study identified and extracted substructures associated with severe Adverse Drug Reactions (ADRs), validating their presence within drug structures through a comparison with a random set of drug structures. Predictions were made for specific molecular structures, and results were validated using clinical evidence from the literature. Results: Optimized molecular fingerprint algorithms and diverse machine-learning models yielded promising outcomes. The Area Under Curve (AUC) value for the fingerprint dataset was obtained at approximately 65%, and integrating it with patient data significantly improved the performance by about 30%. Substructure analysis pinpointed key components linked to severe ADRs, reinforcing the predictive prowess of the model. Predictions for specific molecular structures were corroborated using clinical evidence from the literature, fortifying the credibility of the proposed approach. Conclusion: In conclusion, this research effectively tackles challenges in the early detection and prediction of ADRs by leveraging machine learning algorithms, focusing on drug molecular substructures. The optimized model, incorporating both fingerprint and patient datasets, demonstrated significant improvements in predictive performance. Identifying and validating substructures linked to severe ADRs contribute to the model's reliability. The study's findings are vital for advancing drug safety and laying the groundwork for further strides in predictive modeling within the medical domain.

Adverse Drug Reactions prediction by combining wide & deep learning and POLY2

Accurate prediction of Adverse Drug Reactions (ADRs) holds immense importance in the field of clinical medicine and drug development. The requirement of accurate prediction spans various stages, ranging from drug design and clinical trials to marketing monitoring. The traditional ADR forecasting method has the disadvantage that it requires a lot of computing resources and is not suitable for large-scale forecasting. To address this issue, this study introduces the Wide & Deep model. This model combines the abilities of memorization and generalization to enhance the accuracy of ADR predictions. Additionally, we identify a shortcoming in the wide component of the traditional Wide & Deep model the lack of nonlinear transformation. Therefore, we propose the inclusion of POLY2 in the Wide & Deep model to rectify this shortcoming. By incorporating POLY2, our aim is to retain the models memorization and generalization abilities, leverage the nonlinear relationship between features, and capture the interaction effect between drug chemical substructures for better model performance. To validate our proposed method, we conduct experiments on two datasets: the FDA Adverse Event Reporting System (FAERS) and PubChem. The evaluation metric utilized is the Area Under the Curve (AUC) score, which demonstrates that our method outperforms the original model. The results indicate that by combining POLY2 feature crosses with the Wide & Deep model, we have achieved significant improvements in the prediction of ADRs.

Non-Negative matrix factorization combined with kernel regression for the prediction of adverse drug reaction profiles.

Post-market unexpected Adverse Drug Reactions (ADRs) are associated with significant costs, in both financial burden and human health. Due to the high cost and time required to run clinical trials, there is significant interest in accurate computational methods that can aid in the prediction of ADRs for new drugs. As a machine learning task, ADR prediction is made more challenging due to a high degree of class imbalance and existing methods do not successfully balance the requirement to detect the minority cases (true positives for ADR), as measured by the Area Under the Precision-Recall (AUPR) curve with the ability to separate true positives from true negatives [as measured by the Area Under the Receiver Operating Characteristic (AUROC) curve]. Surprisingly, the performance of most existing methods is worse than a naïve method that attributes ADRs to drugs according to the frequency with which the ADR has been observed over all other drugs. The existing advanced methods applied do not lead to substantial gains in predictive performance. We designed a rigorous evaluation to provide an unbiased estimate of the performance of ADR prediction methods: Nested Cross-Validation and a hold-out set were adopted. Among the existing methods, Kernel Regression (KR) performed best in AUPR but had a disadvantage in AUROC, relative to other methods, including the naïve method. We proposed a novel method that combines non-negative matrix factorization with kernel regression, called VKR. This novel approach matched or exceeded the performance of existing methods, overcoming the weakness of the existing methods. Code and data are available on https://github.com/YezhaoZhong/VKR.

Pattern of central nervous system related adverse drug reactions (ADRs) due to anti-retroviral therapy in HIV infected patients in tertiary care hospitals in North India

Background: Anti-retroviral therapy is associated with myriads of side-effects. Central nervous system related side effects are troublesome and mostly self limiting. Working knowledge of these side effects can improve the prescription habits of physicians. The study aimed pattern of central system related adverse drug reactions (ADRs) due to anti-retroviral therapy in HIV infected patients in tertiary care hospitals in North India. Materials and methods: A prospective observational study was conducted and all HIV positive patients, previously registered and new; attending OPD who encountered ADRs were enrolled in our study irrespective of their age and sex. Data was collected using ADR reporting form issued by Indian Pharmacopoeia Commission. Causality assessment was done by using Naranjo’s Probability Scale. Modified Hartwig severity scale was used to evaluate severity, WHO criteria for seriousness and guidelines of council for international organizations of medical sciences to decide the predictability of ADRs. Results: A total of 250 patients encountered various types of ADRs during the study period. Total number of ADRs recorded was 452. Out of total ADRs (n=452) recorded, 20.35% (n=92) were central system related ADRs. Most common central nervous system related ADR was dizziness. Most central nervous system related ADRs were of moderate type and possible in nature on causality assessment. Conclusion: Relatively overweight, older age and baseline CD4 absolute count below 250 cells/mm3, were predictors for central nervous system related side effects due to anti-retroviral therapy in HIV patients. We suggest that more stringent evaluation and monitoring should be carefully done in all HIV/AIDS patients who will receive and on efavirenz based anti-retroviral therapy, especially if they have the above predictors.

Improvements in adverse drug reaction prediction

This report investigates prediction on adverse drug reactions (ADR) with kernel and imbalance data mechanisms. The hypothesis is that different types of kernel lead to different prediction results, which suggests deciding the best-fit kernel might be a critical way of improving prediction accuracy. Besides, it was also hypothesized that edge cases in real-life setting would cause imbalance in the dataset, thus further causing inaccuracy in prediction. Similarly, attempting to add class weight to various machine learning models could also be a way to improve prediction accuracy. Hence, these hypotheses are being explored in this study.

PADR: Towards Personalized Adverse Drug Reaction Prediction by Modeling Multi-sourced Data.

Predicting adverse drug reactions (ADRs) of drugs is one of the most critical steps in drug development. By pre-estimating the adverse reactions, researchers and drug development companies can greatly prevent the potential ADR risks and tragedies. However, the current ADR prediction methods suffer from several limitations. First, the prediction results are based on pure drug-related information, which makes them impossible to be directly applied for the personalized ADR prediction task. The lack of personalization of models also makes rare adverse events hard to be predicted. Therefore, it is of great interest to develop a new personalized ADR prediction method by introducing additional sources, e.g., patient health records. However, few methods have tried to use additional sources. In the meantime, the variety of different source formats and structures makes this task more challenging. To address the above challenges, we propose a novel personalized multi-sourced-based drug adverse reaction prediction model named pADR. pADR first works on every single source to transform them into proper representations. Next, a hierarchical multi-sourced Transformer is designed to automatically model the interactions between different sources and fuse them together for the final adverse event prediction. Experimental results on a new multi-sourced ADR prediction dataset show that PADR outperforms state-of-the-art drug-based baselines. Moreover, the case and ablation studies also illustrate the effectiveness of our proposed fusion strategies and the reasonableness of each module design.

Application of artificial intelligence and machine learning in early detection of adverse drug reactions (ADRs) and drug-induced toxicity

Adverse drug reactions (ADRs) and drug-induced toxicity are major challenges in drug discovery, threatening patient safety and dramatically increasing healthcare expenditures. Since ADRs and toxicity are not as visible as infectious diseases, the potential consequences are considerable. Early detection of ADRs and drug-induced toxicity is an essential indicator of a drug's viability and safety profile. The introduction of artificial intelligence (AI) and machine learning (ML) approaches has resulted in a paradigm shift in the field of early ADR and toxicity detection. The application of these modern computational methods allows for the rapid, thorough, and precise prediction of probable ADRs and toxicity even before the drug’s practical synthesis as well as preclinical and clinical trials, resulting in more efficient and safer medications with a lesser chance of drug’s withdrawal. This present review offers an in-depth examination of the role of AI and ML in the early detection of ADRs and toxicity, incorporating a wide range of methodologies ranging from data mining to deep learning followed by a list of important databases, modeling algorithms, and software that could be used in modeling and predicting a series of ADRs and toxicity. This review also provides a complete reference to what has been performed and what might be accomplished in the field of AI and ML-based early identification of ADRs and drug-induced toxicity. By shedding light on the capabilities of these technologies, it highlights their enormous potential for revolutionizing drug discovery and improving patient safety.

NA

NA

English

Drugs for medical purpose aims to save a person’s life and improve the quality of life. They enhance one’s mental or physical fitness and treat disease. However, drugs might cause adverse reactions or even unexpected effects on patients, known as Adverse Drug Reactions (ADRs). For preventing ADRs, drug trials were carried out to evaluate and analyze the potential danger in the procedure of drug development. Deep learning (DL) is a type of Machine Learning (ML) in Artificial intelligence (AI) that has developed as a highly effective and promising method that can interrogate and combine diverse biological data kinds to produce new hypotheses. DL is extensively used in drug repurposing and discovery, but its application in ADR prediction employing gene expression data is limited. Therefore, this study introduces a Deer Hunting Optimization Driven Deep Learning Model for Robust Adverse Drug Reaction Recognition and Classification (DHODL-ADRRC) technique. The DHODL-ADRRC technique involves diverse phases of data preprocessing to normalize the data. In addition, the Binary Fire Hawks Optimization (BFHO) based feature selection (FS) method is used to elect an optimum set of features. Moreover, the Attention-based Convolutional Bidirectional Long Short-Term Memory (ACBLSTM) algorithm is utilized for ADR identification. Furthermore, the DHO model is exploited for the tuning process adjusting of the ACBLSTM model, advancing the classification outputs. The simulation output of the DHODL-ADRRC algorithm was investigated on the ADR dataset, and the extensive outputs highlighted the advanced achievement of the DHODL-ADRRC method over other current methods in terms of different measures.

Clinical and pharmacological approaches to personalization of the prescription of oral anticoagulants: clinical cases

Anticoagulants play an important role in reducing complications and mortality associated with thromboembolic disease. For some time, vitamin K antagonists (VKAs) have been the main drugs used for long-term oral anticoagulant therapy, but because of the significant limitations of VKAs over the past decade, pharmacological research has led to the development of new direct acting oral anticoagulants (DOACs). Direct acting oral anticoagulants have a rapid onset of action with peak levels within 2–4 hours and a half-life of about 12 hours, which is much shorter than that of vitamin K antagonists, a more predictable anticoagulant effect, no need for dose selection, routine laboratory monitoring of pharmacodynamic effects, and a lower frequency of clinically significant drug-drug interactions compared with warfarin. But anticoagulants can still cause serious adverse drug reactions (ADRs) in the form of hemorrhagic complications in hospitalized patients, as confirmed in studies. Currently, clinicalpharmacological technologies of personalized medicine such as pharmacogenetic and pharmacokinetic studies are considered as promising approaches to improve the safety of modern pharmacotherapy, allowing the prediction and prevention of various ADRs. In addition, there are emerging studies showing the importance of genetic features of patients in relation to the metabolism of oral anticoagulants, as well as described clinical situations where different gene polymorphisms, could be responsible for changes in the pharmacokinetics of DOACs. This article reviews clinical cases in which pharmacogenetic testing and therapeutic drug monitoring are used to optimize the clinical efficacy and maximum safety of anticoagulant therapy with apixaban and rivaroxaban.

Adaptive dual graph contrastive learning based on heterogeneous signed network for predicting adverse drug reaction

Adverse Drug Reactions (ADRs) resulting from drug combinations can endanger patients' health and life. In this paper, we propose ADGCL, a novel Adaptive Dual Graph Contrastive Learning based on heterogeneous signed networks for predicting ADRs. Specifically, ADGCL first explicitly models positive and negative drug-drug interactions (DDIs) using a heterogeneous signed network to learn semantic-rich drug feature representations. Additionally, ADGCL employs a dual graph contrastive learning strategy including self-supervised contrastive learning and micro-supervised learning to learn high-level features, which significantly improves prediction performance. Furthermore, ADGCL incorporates an adaptive negative sampling method to generate negative samples. Lastly, an encoder based on the implicit graph neural network is designed to capture long-range dependencies in underlying networks, thereby improving the quality of drug feature representation. Comprehensive experimental results on real-world datasets demonstrate that ADGCL is a promising ADR prediction model.

Transfer Learning for Early Prediction of Adverse Drug Reactions: Docetaxel and Alopecia in Breast Cancer as a Case Study.

Transfer Learning (TL) is an approach which has not yet been widely investigated in healthcare, mostly applied in image data. This study outlines a TL pipeline leveraging Individual Case Safety reports (ICSRs) and Electronic Health Records (EHR), applied for the early detection Adverse Drug Reactions (ADR), evaluated using of alopecia and docetaxel on breast cancer patients as use case.

Abstract 3157: Prediction of adverse drug reactions of anti-breast cancer drugs using machine learning

Abstract Chemotherapy is a representative treatment option for cancers, but one of the challenges during the course of cancer treatment is the possible occurrence of adverse drug reactions (ADRs) caused by an anticancer drug. For example, ADRs such as encephalopathy, radiation recall dermatitis, acute cardiogenic shock and febrile neutropenia have been reported for 5-fluorouracil, an anticancer drug used to treat several cancer types. These ADRs may be caused by a single drug or by interactions with other drugs. Unfortunately, it is extremely difficult to predict potential ADRs a priori because the occurrence of ADR depends on the patient’s genetics and health condition as well as a combination of drugs taken together. To address this problem, we use electric health record (EHR) data of breast cancer patients as a proof-of-concept demonstration to develop a machine learning model that predicts ADRs. For this, administrations of breast cancer patients, diagnostic records, and additional personal information (e.g., age and gender) were used as input data. To deal with high sparsity of EHR data, which has categorical information, factorization machine was applied. Factorization machine can effectively capture key information in an sparse environment. The model classifies whether a patient shows one of pre-defined ADRs as an output. The machine learning model developed in this study can be useful for predicting potential ADRs of various drugs by taking into account a patient’s health condition. Citation Format: Kang Seongmo, Suehyun Lee, Hyunah Shin, Hyun Uk Kim. Prediction of adverse drug reactions of anti-breast cancer drugs using machine learning [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 3157.

Adverse drug reactions in an ageing PopulaTion (ADAPT) study: Prevalence and risk factors associated with adverse drug reaction-related hospital admissions in older patients.

Background: Older people experience greater morbidity with a corresponding increase in medication use resulting in a potentially higher risk of adverse drug reactions (ADRs). Objectives: The aim of this study was to; 1) determine the prevalence and characteristics of ADR-related hospital admissions among older patients (≥65years) in Ireland; and 2) identify the risk factors associated with ADR-related hospital admissions. Methods: A cross-sectional study of ADR prevalence in patients aged ≥65years admitted acutely to hospital in Ireland over a 8 month period (November 2016- June 2017). A multifaceted review of each hospital admission was undertaken to assess the likelihood of an ADR being a reason for admission (cause of admission or contributing to admission) in the context of the patient's medication, clinical conditions, comorbidities and investigations. A number of decision aids were applied by two independent reviewers to assess ADR causality, avoidability and severity. A random sample of patients, determined not to have a suspected ADR on screening, were assigned to a non-ADR control group. Multivariable logistic regression was used to assess the association between potential risk factors for ADR-related admissions compared with non-ADR-related admissions. Results: In total, 3,760 hospital admission episodes (in 3,091 patients) were screened and 377 admissions were considered ADR-related (10.0%, 95% CI 9.1%, 11.0%). 219 (58.1%) ADR-related admissions were caused by an ADR, while ADRs contributed to 158 (41.9%) admissions. 268 (71.1%) of all ADR-related admissions were deemed definitely or possibly preventable/avoidable. 350 (92.8%) ADRs were classified as being of moderate severity, with 27 (7.2%) classified as severe. Antithrombotic agents, mainly aspirin and warfarin, were the drugs most frequently associated with ADR-related admissions (gastrointestinal and vascular haemorrhagic disorders). In multivariable analysis, immobility, frailty, having delirium or ulcer disease and taking anticoagulant and antiplatelet medication on admission were significantly associated with an ADR-related hospital admission. Conclusion: One in ten hospital admissions, among those aged 65 + years, were considered ADR-related, with approximately 70% potentially avoidable. Reliable and validated ADR detection and prediction tools are needed to develop prevention strategies.

Detection and Prediction of Adverse Drug Reactions Using Data Driven Modelling with Machine Learning Models

Various pharmacological, genetic and immunity related reasons contribute to the recorded Adverse Drug Reactions (ADR) either directly or indirectly. The genetic factors are not limited to ethnicity, age and demographics, social and economical and gender factors. Such factors are considered to be the secondary reasons, whereas the primary reasons indicate the interactions between drug-drug, drug-protein and protein-protein entities. With advanced algorithms in Machine Learning and Data Science, the prediction of pharmacovigilance has reached greater heights in the recent past. The conventional methods are time-consuming and demands huge intervention from experts and manufacturers. The models built with machine learning models have simplified the analysis and regression models have identified serious adverse drug reactions better than conventional methods. Allergies, organ failures and haemorrhages are considered for this research work, focusing on the parameters such as age, number of medications consumed, number of illnesses affecting the specific patient, dosage, type of medical institution, previous or genetic history of adverse reactions to medicines, type of consumptions and method of medication intake. From the investigative results, elderly patients affected by multiple illness are bound to multiple medicine intake and thus are subjected to serious adverse drug reactions. Owing to the fact, the monitoring period should be shortened and supervised accordingly. Primary responsibility of medical institutions lies in monitoring the previous history of adverse reactions and the new symptoms upon consumption of new medicine. The proposed approach carefully studies the relationships between various factors and computes in a binary logistic model for effective detection and prediction. The outcomes of the proposed model justify the need for additional parameters for a promising accuracy in detection and prediction of ADRs.

A Machine Learning Framework to Predict Adverse Drug Reactions from Electronic Health Records

The advent of medical industry is immense in the recent years, with increasing epidemics, pandemics and endemics, thereby ensuring proper treatment to the patients with affordable medications. Right from the chemical compositions, the manufacturing involves a huge procedural overhead along with the clinical trials and pre-marketing studies. Such medications have to clear stringent norms and policies in order to reach the market, during which, the medication should exhibit compliance to lesser side effects, and increased control over the diseases. Adverse Drug Reactions (ADR) are typically reactions which are unintended and may result in harmful effects over the patients. Multiple models understand the data collected from various sources to monitor the effects of medications in regular intervals, in order to prepare for counter actions. The risk of side effects and adverse drug reactions can be reduced with timely detection of drug to protein interaction and drug to drug interactions. It is a practice of co-prescriptions for addressing multiple medical conditions in elderly people and patients with multiple ailments. Compared to the previous manual techniques, prediction of adverse drug reactions was carried out using machine learning techniques lately. The proposed technique introduces a novel mechanism using regularized logistic regression technique to effectively trace the drug-to-drug interactions. The datasets are considered from openly available sources, and electronically stored information are fed into regression models for finding relevant patterns. Empirical studies applied with necessary cross validation checks and numerous failproof tests deliver promising outcomes in form of drug-ADR targeted profiles for signifying the results of the supposed study. From the investigative results, it is evident that the proposed technique ensures utmost quality and interesting insights for making appropriate biological and protein-drug based decisions.

Immune regulation and emerging roles of noncoding RNAs in Mycobacterium tuberculosis infection.

Tuberculosis, caused by Mycobacterium tuberculosis, engenders an onerous burden on public hygiene. Congenital and adaptive immunity in the human body act as robust defenses against the pathogens. However, in coevolution with humans, this microbe has gained multiple lines of mechanisms to circumvent the immune response to sustain its intracellular persistence and long-term survival inside a host. Moreover, emerging evidence has revealed that this stealthy bacterium can alter the expression of demic noncoding RNAs (ncRNAs), leading to dysregulated biological processes subsequently, which may be the rationale behind the pathogenesis of tuberculosis. Meanwhile, the differential accumulation in clinical samples endows them with the capacity to be indicators in the time of tuberculosis suffering. In this article, we reviewed the nearest insights into the impact of ncRNAs during Mycobacterium tuberculosis infection as realized via immune response modulation and their potential as biomarkers for the diagnosis, drug resistance identification, treatment evaluation, and adverse drug reaction prediction of tuberculosis, aiming to inspire novel and precise therapy development to combat this pathogen in the future.

Develop an ADR prediction system of Chinese herbal injections containing Panax notoginseng saponin: a nested case–control study using machine learning

ObjectiveThis study aimed to develop an adverse drug reactions (ADR) antecedent prediction system using machine learning algorithms to provide the reference for security usage of Chinese herbal injections containing Panax...