Hierarchical multi-label classification applied to medical findings

• Grant number: 09/04029-8
• Support Opportunities: Scholarships in Brazil - Master
• Start date: September 01, 2009
• End date: November 30, 2010
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Maria Carolina Monard
• Grantee: Everton Alvares Cherman
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

This work proposes the use of hierarchical multi-label structures for classifying collections of medical findings which are described in natural language (Portuguese). To this end, it is proposed an extension of a methodology which was idealize to support, through a semi-automatic process, the construction of a table in the attribute-value format from information contained in medical findings. This extension should take into account the text related to diagnosis, which should be transformed into a hierarchical multi-label structure. These extensions will be integrated into the computational tool TP-Discover in order to include the class attribute into the generated attribute-value table so that classifiers can be built using hierarchical multi-label learning algorithms. The proposal will be evaluated using collections of upper digestive endoscopies medical findings. (AU)