Article
Engineering, Chemical
Wenlong Wang, Qilei Liu, Yachao Dong, Jian Du, Qingwei Meng, Lei Zhang
Summary: This article proposes a framework for predicting potential organic reactions using reaction templates and a 2D convolutional neural network (2D CNN) model. The traditional theory-based technologies are not efficient, so this method offers a more efficient approach. By training the 2D CNN models with 605,753 patented reactions and their generated counterparts from the USPTO 1976-2016 database, the models are able to evaluate the likelihood of molecular transformations by learning the feature differences between reactants and products. The classification accuracies of the models for non-trained reactions are 97.881% and 99.593% respectively, and challenging reactions from literature regarding selectivity are correctly predicted. In addition, a visual reaction fingerprint is introduced to provide novel insights into the model interpretability.
Article
Biochemical Research Methods
Jana Schor, Patrick Scheibe, Matthias Bernt, Wibke Busch, Chih Lai, Joerg Hackermueller
Summary: Many chemicals in the environment pose risks if not assessed properly; limitations in computational approaches due to lack of labeled training data is a major challenge.
BRIEFINGS IN BIOINFORMATICS
(2022)
Review
Biochemical Research Methods
Chun Yen Lee, Yi-Ping Phoebe Chen
Summary: Traditional machine learning methods for detecting drug side effects are labor-intensive, while deep learning approaches, with the integration of heterogeneous drug data sources and innovative deployment, can help reduce adverse drug reactions and find replacements for drugs with side effects, as well as diversify drug utilization through drug repurposing.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Computer Science, Interdisciplinary Applications
Sara Santiso, Alicia Perez, Arantza Casillas
Summary: This work focuses on extracting ADRs from EHRs in Spanish using a deep neural network approach. The system shows the ability to handle cross-hospital predictions and has a certain tolerance to external variations. However, errors in entity recognition can lead to a decrease in performance in the ADR detection stage.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2021)
Article
Computer Science, Artificial Intelligence
Humayun Kayesh, Md Saiful Islam, Junhu Wang, Ryoma Ohira, Zhe Wang
Summary: This paper proposes a novel approach for detecting ADRs in Tweets by exploiting cause-effect relationships. By splitting Tweets into different segments, extracting causal features, and applying multi-head self-attention mechanism, ADRs can be effectively detected. Extensive experiments demonstrate the effectiveness of the proposed approach.
Article
Health Care Sciences & Services
Jhih-Yuan Huang, Wei-Po Lee, King-Der Lee
Summary: Social forums provide new channels for constructing predictive models of adverse drug reactions (ADRs) for post-marketing surveillance. However, challenges still exist due to the characteristics of social posts. To address these issues, we performed data analytics from the perspectives of data balance, feature selection, and feature learning, and introduced a deep learning-based approach to enhance predictive performance.
Article
Computer Science, Theory & Methods
Xiongjun Guan, Yongjie Duan, Jianjiang Feng, Jie Zhou
Summary: Skin distortion is a challenge in fingerprint matching, and previous studies have focused on estimating and rectifying the distortion field to improve recognition rate. However, existing rectification methods based on principal component representation are not accurate and sensitive to finger pose. This paper proposes a rectification method using a self-reference based network to directly estimate the dense distortion field of distorted fingerprints, achieving state-of-the-art performance in distortion field estimation and rectified fingerprint matching.
IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY
(2023)
Article
Chemistry, Multidisciplinary
Zhaofeng Ye, Fengling Chen, Jiangyang Zeng, Juntao Gao, Michael Q. Zhang
Summary: This study introduces a new method, ScaffComb, which integrates phenotype information and molecular scaffolds for drug combination screening in large-scale databases, resulting in the discovery of novel drug combinations and the identification of new synergistic mechanisms.
Article
Computer Science, Interdisciplinary Applications
Pratik Joshi, V. Masilamani, Anirban Mukherjee
Summary: Artificial Intelligence (AI) has been increasingly used in drug discovery, particularly in predicting adverse drug reactions (ADRs). This study proposes a novel method based on knowledge graph embedding to predict ADRs by training a custom-made deep neural network model. The results demonstrate that this method outperforms existing approaches in terms of prediction accuracy.
JOURNAL OF BIOMEDICAL INFORMATICS
(2022)
Review
Biochemical Research Methods
Duc Anh Nguyen, Canh Hao Nguyen, Hiroshi Mamitsuka
Summary: This paper provides an overview of ADR studies, focusing on data sources, three main tasks, machine learning methods, and performance comparisons for drug-ADR prediction. The study highlights the importance of utilizing machine learning approaches in analyzing and predicting ADRs, while also acknowledging ongoing challenges and open problems in this field.
BRIEFINGS IN BIOINFORMATICS
(2021)
Article
Toxicology
Minjie Xu, Hongbin Yang, Guixia Liu, Yun Tang, Weihua Li
Summary: This study constructed predictive models for fish toxicity, including local models and a global model. Traditional machine learning methods and deep learning methods were used, and it was found that deep learning had advantages in performance. These models can be used for chemical risk assessment and drug discovery.
JOURNAL OF APPLIED TOXICOLOGY
(2022)
Article
Chemistry, Multidisciplinary
Jingyi Hou, Zhen Dong
Summary: This paper proposes an explainable and generalized chemical reaction representation method to accelerate the evaluation of chemical processes in production. The method incorporates a small amount of expert knowledge and uses a probabilistic data augmentation strategy with contrastive learning to improve model generalization. Experimental results demonstrate that the method outperforms state-of-the-art approaches by pretraining the model with a small-scale dataset annotated with both coarse-level and fine-level labels.
APPLIED SCIENCES-BASEL
(2023)
Article
Chemistry, Multidisciplinary
Akshai P. Sreenivasan, Philip J. Harrison, Wesley Schaal, Damian J. Matuszewski, Kim Kultima, Ola Spjuth
Summary: This paper proposes a method for predicting protein target clusters using deep neural networks. The model is trained on compound-protein and protein-protein interaction data, and achieves accurate predictions.
JOURNAL OF CHEMINFORMATICS
(2022)
Article
Biochemical Research Methods
Yi Cao, Zhen-Qun Yang, Xu-Lu Zhang, Wenqi Fan, Yaowei Wang, Jiajun Shen, Dong-Qing Wei, Qing Li, Xiao-Yong Wei
Summary: The ATC classification is important in drug development and research, but previous methods relied on lab experiments. A new method is proposed in this study, which solely predicts ATC based on molecular structures. The method outperforms previous methods in accuracy and efficiency.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Biochemical Research Methods
Zhaorui Zuo, Penglei Wang, Xiaowei Chen, Li Tian, Hui Ge, Dahong Qian
Summary: A novel deep-learning model integrating gene expression, genetic mutation, and chemical structure was applied to cancer drug sensitivity datasets, showing that combining these features greatly enhances predictive performance.
BMC BIOINFORMATICS
(2021)
Article
Pharmacology & Pharmacy
Lei Wang, Aditi Shendre, Chien-Wei Chiang, Weidan Cao, Xia Ning, Ping Zhang, Pengyue Zhang, Lang Li
Summary: This study systematically screened known PK mechanisms of DDIs with ADEs using a large surveillance database, identifying 149 CYP substrates and 62 inhibitors. Results showed 590 ADEs associated with 2085 PK DDI pairs and 38 individual substrates, with overlapping ADEs across different CYP substrates. Further clinical, population-based, and experimental studies are needed to confirm the findings.
BRITISH JOURNAL OF CLINICAL PHARMACOLOGY
(2022)
Article
Multidisciplinary Sciences
Dongdong Zhang, Samuel Yang, Xiaohui Yuan, Ping Zhang
Summary: An automatic classification method for cardiac arrhythmias using a deep neural network was developed, showing effectiveness and comparing performance across different leads.
Article
Medical Informatics
Qianlong Wen, Ruoqi Liu, Ping Zhang
Summary: This study introduces a novel framework for drug repurposing based on clinical connectivity mapping, which analyzes the therapeutic effects of drugs on diseases using clinical data. Experimental results demonstrate the effectiveness of this framework in evaluating the repurposing potential of 392 drugs for 6 important chronic diseases.
BMC MEDICAL INFORMATICS AND DECISION MAKING
(2021)
Article
Engineering, Biomedical
Ruhan Liu, Liang Ou, Bin Sheng, Pei Hao, Ping Li, Xiaokang Yang, Guangtao Xue, Lei Zhu, Yuyang Luo, Ping Zhang, Po Yang, Huating Li, David Dagan Feng
Summary: The study aims to accurately identify m(6)A modifications using DRS data, proposing a methodology that incorporates mapping and feature extraction, while introducing the MWNB model for detecting RNA modifications. The experimental results demonstrate high classification accuracy.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING
(2022)
Review
Health Care Sciences & Services
Weidan Cao, M. Wesley Milks, Xiaofu Liu, Megan E. Gregory, Daniel Addison, Ping Zhang, Lang Li
Summary: This systematic review examines the definition, role, and implementation of engagement in hypertension-focused mobile health interventions. The review finds that increased engagement is associated with better biomedical outcomes, and tailoring and interactivity are important determinants of engagement. The review suggests a patient-centered engagement framework for hypertension self-management using mHealth technology.
JMIR MHEALTH AND UHEALTH
(2022)
Review
Health Policy & Services
Tasneem Motiwala, Ping Zhang, Megan Gregory, Naleef Fareed, Xia Ning, Kevin Coombes, Gabrielle Kokanos, Courtney Hebert
Summary: This paper discusses the evaluation methods of applied health informatics courses and presents a survey developed to identify the skills and knowledge base of the faculty. The results of this assessment allowed the authors to identify gaps and develop strategies for program expansion.
LEARNING HEALTH SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Thai-Hoang Pham, Changchang Yin, Laxmi Mehta, Xueru Zhang, Ping Zhang
Summary: In the healthcare domain, complication risk profiling is a challenging task due to complex interactions between clinical entities. Existing methods in deep learning lack optimal models, interpretation mechanisms, and fairness solutions. We propose MuViTaNet, a multi-view multi-task network that improves patient representation, generates generalized representations, and promotes fairness. MuViTaNet outperforms existing methods, provides interpretation mechanisms, and F-MuViTaNet effectively mitigates unfairness with negligible impact on accuracy.
KNOWLEDGE AND INFORMATION SYSTEMS
(2023)
Article
Computer Science, Interdisciplinary Applications
Ruhan Liu, Tianqin Wang, Huating Li, Ping Zhang, Jing Li, Xiaokang Yang, Dinggang Shen, Bin Sheng
Summary: Rare diseases are often neglected in research, but machine learning techniques can greatly improve their diagnosis. The study proposes diagnosis-guided multi-to-mono modal generation networks (TMM-Nets) to transfer data from multiple sources to a single modality for structured diagnostic data. Applying TMM-Nets to diagnose lupus retinopathy (LR-SLE), it demonstrates the potential of using transfer learning based on fundus lesion similarity for rare diseases.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2023)
Article
Computer Science, Artificial Intelligence
Ruoqi Liu, Katherine M. Hunold, Jeffrey M. Caterino, Ping Zhang
Summary: Timely antibiotic treatment is crucial for sepsis management, but confounders in the data affect its accuracy. Liu et al. propose a method called T4, which predicts individual treatment effects with increased accuracy and uncertainty estimation. T4 encodes temporal and static variables to estimate treatment effects and adjusts confounding through mini-batch balancing matching.
NATURE MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Seungyeon Lee, Changchang Yin, Ping Zhang
Summary: This study proposes a new method to address the issue of performance decay in AI models for clinical risk prediction. By reweighting patients and using a specific loss function, our model efficiently mitigates distribution shifts and improves prediction performance in the post-shift environment.
Article
Computer Science, Artificial Intelligence
Thai-Hoang Pham, Yue Qiu, Jiahui Liu, Steven Zimmer, Eric O'Neill, Lei Xie, Ping Zhang
Summary: Chemical-induced gene expression profiles provide crucial information for drug discovery, but current methods are limited by time and cost constraints. In this study, a novel method called CIGER is proposed to predict overall rankings in gene expression profiles and applied to treatment screening of drug-resistant cancer.
Proceedings Paper
Computer Science, Artificial Intelligence
Biplob Biswas, Thai-Hoang Pham, Ping Zhang
Summary: ICD coding procedure is crucial for the medical billing system, but manual assignment of ICD codes may result in errors, making automation an important task. A transformer-based architecture with code-wise attention mechanism is proposed to automate the process, outperforming other baselines and providing more insights to support clinical decisions.
ARTIFICIAL INTELLIGENCE IN MEDICINE (AIME 2021)
(2021)
Article
Computer Science, Artificial Intelligence
Dongdong Zhang, Changchang Yin, Katherine M. Hunold, Xiaoqian Jiang, Jeffrey M. Caterino, Ping Zhang
Summary: The study introduced a LSTM-based model to predict the onset of sepsis, achieving high prediction accuracy. The model utilized event embedding, time encoding, attention mechanism, and global max pooling techniques to improve performance and interpret results effectively.
Article
Computer Science, Interdisciplinary Applications
Yuanfang Guan, Hongyang Li, Daiyao Yi, Dongdong Zhang, Changchang Yin, Keyu Li, Ping Zhang
Summary: The article introduces a statistical modeling method that can be applied to various regression learning algorithms, including deep learning, with empirical advantages in survival prediction problems. The authors demonstrate the method's application in traditional survival problems as well as different types of regression learning algorithms, and showcase its use in clinical informatic data, pathological images, and the hardware industry.
NATURE COMPUTATIONAL SCIENCE
(2021)
Article
Computer Science, Artificial Intelligence
Thai-Hoang Pham, Yue Qiu, Jucheng Zeng, Lei Xie, Ping Zhang
Summary: In drug discovery and repurposing, the Deep Chemical Expression graph neural network is developed to predict chemical-induced gene expression profiles and applied to drug repurposing for COVID-19 treatments. Experimental results show superior performance and support for downstream classification tasks.
NATURE MACHINE INTELLIGENCE
(2021)