<XML><RECORDS><RECORD><REFERENCE_TYPE>3</REFERENCE_TYPE><REFNUM>8031</REFNUM><AUTHORS><AUTHOR>Rogers,S.</AUTHOR><AUTHOR>Girolami,M.</AUTHOR><AUTHOR>Krebs,R.</AUTHOR><AUTHOR>Mischak,H.</AUTHOR></AUTHORS><YEAR>2005</YEAR><TITLE>Disease Classification with Capillary Electrophoresis:Mass Spectrometry</TITLE><PLACE_PUBLISHED>International Conference for Advances in Pattern Recognition, Bath 2005</PLACE_PUBLISHED><PUBLISHER>N/A</PUBLISHER><LABEL>Rogers:2005:8031</LABEL><ABSTRACT>We investigate the possibility of using pattern recognition techniques to classify various disease types using data produced by a new form of rapid Mass Spectrometry. The data format has several advantages over other high-throughput technologies and as such could become a useful diagnostic tool. We investigate the binary and multi-class performances obtained using standard classifiers as the number of features is varied and conclude that there is potential in this technique and suggest research directions that would improve performance. </ABSTRACT><NOTES>Copyright Springer-Verlag (http://www.springer.de/comp/lncs/index.html)</NOTES></RECORD></RECORDS></XML>