Abstract
Clinical trials are human research studies that aim to evaluate a medical, surgical, or behavioral intervention that is critical to the advancement of medical science. The majority of clinical trials fail because recruitment goals are not met. This issue necessitates the incorporation of automated systems capable of matching patients to ongoing clinical trials. This paper summarizes our participation in the TREC 2021 clinical trials track, which provided all participants with a 5–10 sentence patient description and a clinical trials database from Clinical-Trials. gov for matching. Our submission consists of a variety of retrieval techniques, including BM25, entity recognition, BERT, and others. The results show that a simple BM25 ranking algorithm could outperform neural network-based models, mainly due to the absence of quality training data.
Sameh Frihat
Researcher in the first cohort
My research interests include Information Retrieval, Natural Language Processing, Machine Learning, and Explainable AI.
Norbert Fuhr
Principal Investigator
My research interests include Information Retrieval, Natural Language Processing and Computer Science.