Related Experiment Videos
Evaluating a normalized conceptual representation produced from natural language patient discharge summaries
P Zweigenbaum1, J Bouaud, B Bachimont
1DIAM, Service d'Informatique Médicale, Assistance Publique, Hôpitaux de Paris. pz@biomath.jussieu.fr
Summary
The Menelas project developed a method to evaluate conceptual representations from patient summaries. This approach simplifies quality assessment for natural language processing in healthcare.
Area of Science:
- Natural Language Processing
- Medical Informatics
- Health Data Science
Background:
- Patient discharge summaries contain complex clinical information.
- Automated conceptual representation extraction is challenging.
- Evaluating the quality of such representations requires robust methods.
Purpose of the Study:
- To present a method for measuring the quality of conceptual representations generated from patient discharge summaries.
- To apply this evaluation method to the French Menelas prototype.
- To examine the proposed method within the Friedman and Hripcsak framework.
Main Methods:
- Development of a quality measurement method for conceptual representations.
- Application of the method to the Menelas French prototype.
- Analysis of the evaluation method using the Friedman and Hripcsak framework.
- Proposal of conditions to reduce evaluation workload.
Main Results:
- A functional method for evaluating the quality of normalized conceptual representations was developed and applied.
- The evaluation framework was tested on a real-world prototype.
- Insights into reducing the effort required for quality assessment were gained.
Conclusions:
- The developed method provides a viable approach for assessing the quality of conceptual representations from clinical text.
- The study contributes to the evaluation of natural language processing systems in healthcare.
- Proposed conditions can streamline the quality assurance process for similar systems.