BD.20.014 – From Paper to Practice: Improving Patient Outcomes by Informing Clinical Guides with Insights Generated from Big Medical Literature Collections

Route: Creating Value through responsible access to and use of big data

Cluster question: 094 How do we improve the quality of health care as much as possible while keeping it affordable?

Today, medical literature grows by a million new papers per year. New results must be put into action quickly in order to improve patient outcomes. Moving from paper to practice requires facing the three Vs of Big Data: The Volume of papers, the Velocity at which they are published, and the Variety of quality of the papers. New AI technologies have clear potential to help. However, currently AI can filter textual content in only a limited way. Systems that sort papers by topic miss the finer-grained details that reflect the quality of the research and its relevance (i.e., importance for clinical practice). In short, state-of-the-art AI is not able to analyze and retrieve papers meeting the complex criteria used by the creators of clinical guidelines to evaluate newly published research results. In this project, we would like to develop advanced natural language processing and information retrieval techniques that can be closely steered by medical professionals. Our objective is to shorten the path between research results and improved patient care by developing AI that can analyze the stream of recent scientific publications, and alert guideline creators and those who need to take action (e.g., pharmaceutical companies in case of adverse events) to new papers that are important because of both their quality and relevance. These two criteria trade off in ways that are beyond the capacity of existing summarization systems and search engines. Out-of-date clinical knowledge leads to low quality healthcare, a waste of public money, and is harmful, even resulting in the death of patients. Recently, the Federation of Medical Specialists (https://www.demedischspecialist.nl/nieuws/richtlijnen-voortaan-sneller-actueel) has turned its attention to this problem. This project will develop AI approaches to reduce the time and effort needed to derive insights from medical literature that improve clinical guidelines.

Keywords

Big Data, clinical guidelines, Complex information needs, Deep Learning, Information Extraction, Information Retrieval, medical literature, NLP

Other organisations

CWZ

Submitter

Organisation Center for Language Studies, Radboud University (RU)
Name Prof. dr. M.A. (Martha) Larson
E-mail m.larson@let.ru.nl
Website https://www.ru.nl/cls