Literature based Clinical Decision Support: Searching Articles According to Medical Needs

Physicians often seek helps from scientific medical literature for better diagnoses and treatment, especially when dealing with intractable cases. The task is becoming more and more difficult as thousands of new articles are published every month. It’s quite challenging and helpful to assist physicians locating highly relevant articles in limited time. In this study, targeting at fulfilling the task of the TREC-2016 Clinical Decision Support track, we explore effective methods to search medical articles according to physicians’ information needs. The best solution we obtained takes the following strategies: using Medical Text Indexer for query expansion, concentrating on articles with abstracts, tagging articles according to different information needs, and combining results of multiple models.