MACHINE LEARNING TECHNIQUES IN EARLY SCREENING FOR GASTRIC AND ESOPHAGEAL CANCER

A database on 2692 dyspeptic patients over the age of 40 was establish ed, consisting of 73 epidemiological and clinical variables. A tree-ba sed machine learning algorithm (PREDICTOR) was applied to this databas e, in order to attempt to find rules which would classify patients int o 2 groups, i.e., those suffering from gastric or oesophageal cancer, and the remainder. The results were encouraging. The cross-validated c lassification performance figures showed that, by classifying 61.3% of the patients as high risk, a sensitivity of 94.9% and a specificity o f 39.8% could be achieved. It is planned to construct an expert system based on the rules produced by the machine learning algorithm, in ord er to provide preliminary screening for cancer in dyspeptic patients.