Positional Language Models for Clinical Information Retrieval

Proceedings of the 2010 Conference on Empirical Methods in Natural Language Processing, pages 108–115, MIT, Massachusetts, USA, 9-11 October 2010. c⃝2010 Association for Computational Linguistics Positional Language Models for Clinical Information Retrieval Florian Boudin DIRO, Universit´e de Montr´eal CP. 6128, succ. Centre-ville H3C 3J7 Montr´eal, Canada boudinfl@iro.umontreal.ca Jian-Yun Nie DIRO, Universit´e de Montr´eal CP. 6128, succ. Centre-ville H3C 3J7 Montr´eal, Canada nie@iro.umontreal.ca Martin Dawes Department of Family Medicine McGill University, 515 Pine Ave H2W 1S4 Montr´eal, Canada martin.dawes@mcgill.ca Abstract The PECO framework is a knowledge repre- sentation for formulating clinical questions. Queries are decomposed into four aspects, which are Patient-Problem (P), Exposure (E), Comparison (C) and Outcome (O). However, no test collection is available to evaluate such framework in information retrieval. In this work, we first present the construction of a large test collection extracted from system- atic literature reviews. We then describe an analysis of the distribution of PECO elements throughout the relevant documents and pro- pose a language modeling approach that uses these distributions as a weighting strategy. In our experiments carried out on a collection of 1.5 million documents and 423 queries, our method was found to lead to an improvement of 28% in MAP and 50% in P@5, as com- pared to the state-of-the-art method. 1 Introduction In recent years, the volume of health and biomedi- cal literature available in electronic form has grown exponentially. MEDLINE, the authoritative reposi- tory of citations from the medical and bio-medical domain, contains more than 18 million citations. Searching for clinically relevant information within this large amount of data is a difficult task that med- ical professionals are often unable to complete in a timely manner. A better access to clinical evidence represents a high impact application for physicians. Evidence-Based Medicine (EBM) is a widely ac- cepted paradigm for medical practice (Sackett et al., 1996). EBM is defined as the conscientious, explicit and judicious use of current best evidence in making decisions about patient care. Practice EBM means integrating individual clinical expertise with the best available external clinical evidence from systematic research. It involves tracking down the best evi- dence from randomized trials or meta-analyses with which to answer clinical questions. Richardson et al. (1995) identified the following four aspects as the key elements of a well-built clinical question: • Patient-problem: what are the patient charac- teristics (e.g. age range, gender, etc.)? What is the primary condition or disease? • Exposure-intervention: what is the main in- tervention (e.g. drug, treatment, duration, etc.)? • Comparison: what is the exposure compared to (e.g. placebo, another drug, etc.)? • Outcome: what are the clinical outcomes (e.g. healing, morbidity, side effects, etc.)? These elements are known as the PECO elements. Physicians are educated to formulate their clinical questions in respect to this structure. For example, in the following question: “In patients of all ages with Parkinson’s disease, does a Treadmill training com- pared to no training allows to increase the walking distance?” one can identify the following elements: • P: Patients of all ages …

Content truncated to 500 words