Go to DBLP homepageFrom Handwriting Analysis to Alzheimer's Disease Prediction: An Experimental Comparison of Classifier Combination Methods
Abstract: In recent years, there has been a growing scientific interest in the study of effective methods for the early diagnosis of Alzheimer’s disease. In this area, it is generally agreed that handwriting analysis can provide very promising contributions: the act of writing is the result of a complex process that involves various cognitive functions, including memory, language, and other executive functions, all or partially affected in the early stages of Alzheimer’s disease. Based on these considerations, various handwriting tasks have been proposed and designed to highlight the different abilities that the onset of the disease could compromise, and various classification systems have been developed that use the information derived from such data. It is useful to remark that the results of these tasks should be analyzed jointly since, as mentioned before, the first symptoms of the disease can concern different cognitive abilities of the subjects involved: this explains the importance of applying classifier combination techniques, which allow the responses provided by the individual classifiers on each single task, to be combined for improving the overall classification accuracy. In this framework, our study aims to conduct a large set of experiments to compare the results of combining techniques for diagnosing Alzheimer’s disease. The data used in the experiments relates to a set of 174 subjects, including 89 patients diagnosed with Alzheimer’s disease and 85 healthy controls, to whom a writing test consisting of 34 handwriting tasks was administered. The results obtained allowed us, on the one hand, to evaluate the performance increase obtained with the different combining techniques and, on the other, to characterize the contribution of the different handwriting tasks in the diagnosis of Alzheimer’s disease.
Loading