A Counterpropagation network based system for screening of mild cognitive Impairment

Abstract

Alzheimer's Disease (AD) and other dementias are one of the public health challenges mainly because of the relationship between population longevity and the increase of the pathology incidence. Furthermore, first symptoms appear several years after beginning of the disease and the progression of the cognitive decline rises over time. Therefore, it is necessary to accomplish diagnosis at the earliest possible stage, since the subject shows a slight impairment in some cognitive function. The detection of this state, named Mild Cognitive Impairment (MCI), is a complex task in medicine. The difficult distinction is between normal ageing and MCI rather than between MCI and AD. In this paper, we propose a CPN based system and a scheme of data fusion to aid MCI diagnosis. We present our preliminary results on MCI detection, using as dataset structure a simple combination of cognitive and functional measurements and the educational level of patients, gathered during clinical consultations. We have tackled an imbalanced classification problem developing a novel extended over-sampling method, SNEOM. Finally, we also performed a comparative study between our intelligent clinical decision system and a clinical expert, revealing the high level of performance of our proposal.