Related Experiment Video
Updated: Apr 25, 2026
03:14
Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness
Published on: December 6, 2024
1.3K
Exploring the application of deep learning techniques on medical text corpora
José Antonio Minarro-Giménez1, Oscar Marín-Alonso1, Matthias Samwald1
1Section for Medical Expert and Knowledge-Based Systems, Medical University of Vienna, Vienna, Austria.
Studies in Health Technology and Informatics
|August 28, 2014
Summary
This study explores using deep learning (word2vec) to extract pharmaceutical properties from medical texts. While results were mixed, it highlights potential for improving medical knowledge accessibility.
Area of Science:
- Biomedical Informatics
- Natural Language Processing
- Artificial Intelligence
Background:
- The increasing volume of biomedical literature presents challenges in information retrieval.
- Efficiently accessing and understanding medical knowledge is crucial for research and clinical practice.
Purpose of the Study:
- To evaluate the effectiveness of the word2vec deep learning toolkit for extracting pharmaceutical properties from unstructured medical text.
- To assess the potential of word2vec in enhancing medical knowledge accessibility and complementing existing ontologies.
Main Methods:
- Applied the word2vec deep learning model to mid-sized medical text corpora.
- Identified pharmaceutical properties such as disease treatment and physiological effects.
- Compared identified relationships with the National Drug File - Reference Terminology (NDF-RT) ontology.
Main Results:
- The efficiency of word2vec in identifying pharmaceutical properties yielded mixed results.
- The study identified specific relationships between drugs and diseases or physiological processes.
- Comparison with the NDF-RT ontology provided insights into the model's performance.
Conclusions:
- Deep learning technologies, including word2vec, show promise for advancing medical information retrieval.
- Further research is needed to optimize deep learning applications in this domain.
- Word2vec can potentially complement curated knowledge within medical ontologies and taxonomies.
