Prescription information and adverse drug reactions (ADR) are two components of detailed medication instructions that can benefit many aspects of clinical research. Automatic extraction of this information from free-text narratives via Information Extraction (IE) can open it up to downstream uses. IE is commonly tackled by supervised Natural Language Processing (NLP) systems which rely on annotated training data. However, training data generation is manual, time-consuming, and labor-intensive. It is desirable to develop automatic methods for augmenting manually labeled data. We propose pseudo-data generation as one such automatic method. Pseudo-data are synthetic data generated by combining elements of existing labeled data. We propose and evaluate two sets of pseudo-data generation methods: knowledge-driven methods based on gazetteers and data-driven methods based on deep learning. We use the resulting pseudo-data to improve medication and ADR extraction. Data-driven pseudo-data are suitable for concept categories with high semantic regularities and short textual spans. Knowledge-driven pseudo-data are effective for concept categories with longer textual spans, assuming the knowledge base offers good coverage of these concepts. Combining the knowledge- and data-driven pseudo-data achieves significant performance improvement on medication names and ADRs over baselines limited to the use of available labeled data.
Download full-text PDF |
Source |
|---|---|
| http://dx.doi.org/10.3233/SHTI190249 | DOI Listing |
Publication Analysis
Top Keywords
Similar Publications
International norms for adult handgrip strength: A systematic review of data on 2.4 million adults aged 20 to 100+ years from 69 countries and regions.
J Sport Health Sci
November 2024
Alliance for Research in Exercise, Nutrition and Activity (ARENA), Allied Health and Human Performance, University of South Australia, Adelaide, SA, 5000, Australia.
Article Synopsis
- Muscular strength, particularly measured by handgrip strength (HGS), is an important indicator of health and a predictor of age-related diseases, but no international benchmarks exist for HGS across different ages and sexes.
- The study systematically analyzed data from over 2.4 million adults across 69 countries to establish sex- and age-specific norms for HGS, identifying a peak in strength between ages 30 and 39 before a gradual decline.
- Findings revealed that while absolute and body size-normalized HGS improves slightly in early adulthood, a more significant drop occurs from middle to late adulthood, with males generally experiencing a faster decline than females.
Article Synopsis
- Continual learning (CL) focuses on learning new tasks while retaining knowledge from previous tasks, but traditional methods often struggle due to the lack of available raw data because of copyright and privacy issues.
- This paper introduces two innovative settings for CL: data-efficient CL (DECL-APIs) and data-free CL (DFCL-APIs), which utilize machine learning APIs to train models with little or no raw data, facing unique challenges like incomplete data and model variability.
- A new cooperative continual distillation learning framework is proposed, using adversarial training with generators to create synthetic data from APIs, enabling effective knowledge transfer to a CL model while implementing a regularization term to prevent forgetting previous knowledge; results show that this approach can perform compar
Article Synopsis
- Computer-aided diagnosis (CAD) systems enhance pathologists' ability to accurately and consistently diagnose cancers, but models trained on data from a single center face challenges in generalizing to different medical environments.
- The proposed solution, SSL-FT-BT, utilizes a self-supervised federated learning framework and generates pseudo-histopathological images from each center, allowing for enhanced model training without compromising patient privacy.
- Experimental results demonstrate that this approach improves both diagnostic accuracy and the ability to generalize across different datasets, showing its effectiveness in real-world applications.
Hybrid Deep Learning Approach for Accurate Tumor Detection in Medical Imaging Data.
Diagnostics (Basel)
March 2023
Department of Computer Engineering, Haliç University, Istanbul 34394, Turkey.
The automated extraction of critical information from electronic medical records, such as oncological medical events, has become increasingly important with the widespread use of electronic health records. However, extracting tumor-related medical events can be challenging due to their unique characteristics. To address this difficulty, we propose a novel approach that utilizes Generative Adversarial Networks (GANs) for data augmentation and pseudo-data generation algorithms to improve the model's transfer learning skills for various tumor-related medical events.
View Article and Find Full Text PDFApplication of machine learning techniques to the analysis and prediction of drug pharmacokinetics.
J Control Release
December 2022
Department of Drug Delivery Research, Graduate School of Pharmaceutical Sciences, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan; Department of Applied Pharmacy and Pharmacokinetics, Graduate School of Pharmaceutical Sciences, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan. Electronic address:
In this review, we describe the current status and challenges in applying machine-learning techniques to the analysis and prediction of pharmacokinetic data. The theory of pharmacokinetics has been developed over decades on the basis of physiology and reaction kinetics. Mathematical models allow the reduction of pharmacokinetic data to parameter values, giving insight and understanding into ADME processes and predicting the outcome of different dosing scenarios.
View Article and Find Full Text PDFWant AI Summaries of new PubMed Abstracts delivered to your In-box?
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!