Abstract
The subfield of machine learning focusing on naturally spoken or written language by humans is called natural language processing (NLP). A major task for NLP is transforming free text into structured data and linking text to ontologies or terminologies. Use cases of NLP in the clinical domain are the extraction of diagnoses and staging information, patient matching for clinical trials and the extraction of progress and outcome data from clinical notes. Clinicians can contribute to an NLP system by labeling data or by providing input for rules. NLP approaches include rule-based, traditional machine-learning approaches and neural or deep-learning approaches. With the introduction of new deep-learning models and improved techniques, such as the transformer model, there have been striking performance improvements across a wide range of NLP tasks. It is likely that NLP systems will become more and more widespread in clinical practice in the next years.
Original language | English |
---|---|
Title of host publication | Machine Learning and Artificial Intelligence in Radiation Oncology: A Guide for Clinicians |
Editors | John Kang, Tim Rattay, Barry S. Rosenstein |
Publisher | Elsevier |
Chapter | 6 |
Pages | 137-161 |
Number of pages | 25 |
ISBN (Electronic) | 9780128220009 |
DOIs | |
Publication status | Published - 1 Jan 2023 |
Keywords
- NLP
- Ontologies
- Structured data
- Text
- Transfer learning
- Transformers
- Word embeddings