Similarity Metrics for Clustering PubMed Abstracts for Evidence Based Medicine

Similarity Metrics for Clustering PubMed Abstracts for Evidence Based Medicine Hamed Hassanzadeh† Diego Moll´a♦ Tudor Groza‡ Anthony Nguyen♮ Jane Hunter† †School of ITEE, The University of Queensland, Brisbane, QLD, Australia ♦Department of Computing, Macquarie University, Sydney, NSW, Australia ‡Garvan Institute of Medical Research, Darlinghurst, NSW, Australia ♮The Australian e-Health Research Centre, CSIRO, Brisbane, QLD, Australia h.hassanzadeh@uq.edu.au, diego.molla-aliod@mq.edu.au t.groza@garvan.org.au, anthony.nguyen@csiro.au, jane@itee.uq.edu.au Abstract We present a clustering approach for doc- uments returned by a PubMed search, which enable the organisation of evi- dence underpinning clinical recommenda- tions for Evidence Based Medicine. Our approach uses a combination of document similarity metrics, which are fed to an ag- glomerative hierarchical clusterer. These metrics quantify the similarity of pub- lished abstracts from syntactic, semantic, and statistical perspectives. Several evalu- ations have been performed, including: an evaluation that uses ideal documents as se- lected and clustered by clinical experts; a method that maps the output of PubMed to the ideal clusters annotated by the experts; and an alternative evaluation that uses the manual clustering of abstracts. The results of using our similarity metrics approach shows an improvement over K-means and hierarchical clustering methods using TF- IDF. 1 Introduction Evidence Based Medicine (EBM) is about indi- vidual patients care and providing the best treat- ments using the best available evidence. The mo- tivation of EBM is that clinicians would be able to make more judicious decisions if they had ac- cess to up-to-date clinical evidence relevant to the case at hand. This evidence can be found in schol- arly publications available in repositories such as PubMed1. The volume of available publications is enormous and expanding. PubMed repository, for example, indexes over 24 million abstracts. As 1www.ncbi.nlm.nih.gov/pubmed a result, methods are required to present relevant recommendations to the clinician in a manner that highlights the clinical evidence and its quality. The EBMSummariser corpus (Moll´a and Santiago-martinez, 2011) is a collection of evidence-based recommendations published in the Clinical Inquiries column of the Journal of Family Practice2, together with the abstracts of publications that provide evidence for the recommendations. Visual inspection of the EBMSummariser corpus suggests that a com- bination of information retrieval, clustering and multi-document summarisation would be useful to present the clinical recommendations and the supporting evidence to the clinician. Figure 1 shows the title (question) and ab- stract (answer) associated with one recommenda- tion (Mounsey and Henry, 2009) of the EBM- Summariser corpus. The figure shows three main recommendations for treatments to hemorrhoids. Each treatment is briefly presented, and the qual- ity of each recommendation is graded (A, B, C) ac- cording to the Strength of Recommendation Tax- onomy (SORT) (Ebell et al., 2004). Following the abstract of the three recommendations (not shown in Figure 1), the main text provides the details of the main evidence supporting each treatment, together with the references of relevant publica- tions. A reference may be used for recommending several of the treatments listed in the recommen- dations. Each recommendation is treated in this study as a cluster of references for evaluation …

Content truncated to 500 words