WisPerMed Text at TREC Clinical Trials Track 2021

Abstract

This paper describes the submissions of the WisPerMed Text group to the TREC Clinical Trials Track 2021. It aims to overcome the problems in patient recruitment that often lead to delays or even discontinuation of clinical trials. The focus here is finding methods to improve the process of matching patient case descriptions to eligible clinical trials. For this purpose, different systems were proposed and tested to rank the trials for each patient topic. These systems utilize methods such as transformer-based models, BM25 and keyword extraction. Additionally, Unified Medical Language System (UMLS) was used in an attempt to find relevancy between patient topics and clinical trials based on biomedical concepts. The results obtained showed that the BM25 model based on keyword extraction performed the best out of all our submissions.

Publication
In 30th Text REtrieval Conference
Henning Schäfer
Henning Schäfer
Researcher in the first cohort

My research interests include Deep Learning, Computer Vision, Radiomics, and Explainable AI.

Ahmad Idrissi-Yaghir
Ahmad Idrissi-Yaghir
Researcher in the first cohort

My research interests include Deep Learning, Natural Language Processing, and Information Retrieval.

Wolfgang Galetzka
Wolfgang Galetzka
Researcher in the first cohort

My research interests include Deep Learning, Computer Vision, Radiomics, and Explainable AI.

Christoph M. Friedrich
Christoph M. Friedrich
Co-Speaker

My research interests include Deep Learning, Computer Vision, Radiomics, and Explainable AI.

Next