A Hybrid Approach to Clinical Question Answering
Abstract
In this paper, we describe our clinical question answering system developed and submitted for the Text Retrieval Conference (TREC 2014) Clinical Decision Support (CDS) track. The task for this track was to retrieve relevant biomedical articles to answer generic clinical questions about medical case reports. As part of our maiden participation in TREC, we submitted a single run using a hybrid Natural Language Processing (NLP)-driven approach to accomplish the given task. Evaluation results showed that our clinical question answering system achieved the best scores in two of eight dual-judged topics: #5 and 27, and performed relatively better compared to the median scores for topics: #13, 18, 19, 22, and 23.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2014
- Accession Number
- ADA618772
Entities
People
- Joey Liu
- Oladimeji Farri
- Sadid A. Hasan
- Xianshu Zhu
- Yao Dong