Multiple hierarchical classification of free-text clinical guidelines

Manual classification of free-text documents within a predefined hierarchy is highly time consuming. This is especially true for clinical guidelines, which are often indexed by multiple indices. We present an approach to automated classification of clinical guidelines into predefined hierarchical conceptual categories, based on supervised learning. The hierarchical structure of the concepts is exploited, in the training stage, in order to overcome the low number of training examples. The classification task is broken into a set of intermediate classifications, throughout the hierarchy. We use a vector-space representation in order to represent each guideline and the hierarchical concepts; classification is based on applying a similarity function at each level. Several evaluation measures are suggested, based on the intended use of the hierarchy; these measures correspond to different parameters of the classification algorithm. Preliminary results are encouraging; a formal evaluation, ...

• Type: Journal
• Year: 2006
• Where: ARTMED
• Authors: Robert Moskovitch, Shiva Cohen-Kashi, Uzi Dror, Iftah Levy, Amit Maimon, Yuval Shahar
• Keywords: ARTMED 2006, Clinical Guidelines, Hierarchical Conceptual Categories, Manual Classification