Using artificial neural networks in clinical neuropsychology: High performance in mild cognitive impairment and Alzheimer's disease

Abstract

Mild cognitive impairment (MCI) is a transitional state between normal aging and Alzheimer disease (AD). Artificial neural networks (ANNs) are computational tools that can provide valuable support to clinical decision making, classification, and prediction of cognitive functioning. The aims of this study were to develop, train, and explore and develop the ability of ANNs to differentiate MCI and AD, and to study the relevant variables in MCI and AD diagnosis. The sample consisted of 346 controls and 79 MCI and 97 AD patients. A linear discriminant analysis (LDA) and ANNs with 12 input neurons (10 subtests of a neuropsychological test, the abbreviated Barcelona Test; age; and education), 4 hidden neurons, and output neuron (diagnosis) were used to classify the patients. The ANNs were superior to LDA in its ability to classify correctly patients (100–98.33% vs. 96.4–80%, respectively) and showed better predictive performance. Semantic fluency, working and episodic memory and education showed up as the most significant and sensitive variables for classification. Our results indicate that ANNs have an excellent capacity to discriminate MCI and AD patients from healthy controls. These findings provide evidence that ANNs can be a useful tool for the analysis of neuropsychological profiles related to clinical syndromes.