BackgroundDementia and cognitive impairment associated with aging are a major medical and social concern. Neuropsychological testing is a key element in the diagnostic procedures of Mild Cognitive Impairment (MCI), but has presently a limited value in the prediction of progression to dementia. We advance the hypothesis that newer statistical classification methods derived from data mining and machine learning methods like Neural Networks, Support Vector Machines and Random Forests can improve accuracy, sensitivity and specificity of predictions obtained from neuropsychological testing. Seven non parametric classifiers derived from data mining methods (Multilayer Perceptrons Neural Networks, Radial Basis Function Neural Networks, Support Vector Machines, CART, CHAID and QUEST Classification Trees and Random Forests) were compared to three traditional classifiers (Linear Discriminant Analysis, Quadratic Discriminant Analysis and Logistic Regression) in terms of overall classification accuracy, specificity, sensitivity, Area under the ROC curve and Press'Q. Model predictors were 10 neuropsychological tests currently used in the diagnosis of dementia. Statistical distributions of classification parameters obtained from a 5-fold cross-validation were compared using the Friedman's nonparametric test.ResultsPress' Q test showed that all classifiers performed better than chance alone (p < 0.05). Support Vector Machines showed the larger overall classification accuracy (Median (Me) = 0.76) an area under the ROC (Me = 0.90). However this method showed high specificity (Me = 1.0) but low sensitivity (Me = 0.3). Random Forest ranked second in overall accuracy (Me = 0.73) with high area under the ROC (Me = 0.73) specificity (Me = 0.73) and sensitivity (Me = 0.64). Linear Discriminant Analysis also showed acceptable overall accuracy (Me = 0.66), with acceptable area under the ROC (Me = 0.72) specificity (Me = 0.66) and sensitivity (Me = 0.64). The remaining classifiers showed overall classification accuracy above a median value of 0.63, but for most sensitivity was around or even lower than a median value of 0.5.ConclusionsWhen taking into account sensitivity, specificity and overall classification accuracy Random Forests and Linear Discriminant analysis rank first among all the classifiers tested in prediction of dementia using several neuropsychological tests. These methods may be used to improve accuracy, sensitivity and specificity of Dementia predictions from neuropsychological testing.

To investigate the cognitive capacities of a cohort of ischaemic or haemorrhagic stroke survivors and to identify the clinical determinants of post-stroke cognitive impairment, we evaluated 237 patients admitted to a Stroke Unit (mean age 59; SD=12.7). Three months after stroke, patients were submitted to a neuropsychological evaluation that included the Mini-Mental State Examination (MMSE), a complementary battery to assess specific cognitive domains, the Hamilton Depression Rating Scale (HDRS) and the Blessed Dementia Scale (BDS). Disturbed performance on at least one domain was detected on 131 (55%) patients: 27% had cognitive deficits other than memory, 7% had focal memory deficit, 9% had memory and other cognitive deficits and 6% had dementia. Dementia was associated with female gender (P=0.01), older age (P=0.01) and lower education level (P=0.04). Patients with memory deficits were older (P=0.01) with lower educational level (P=0.08) and more left sided lesions (P=0.02) than patients without memory deficits. In this middle aged stroke survivors cognitive impairment was common 3 months after stroke, while dementia was infrequent.

Background: Quality of life (QoL) is affected in patients with dementia, but it is not clear whether it is already disturbed in more initial phases of cognitive decline, like Mild Cognitive Impairment (MCI). Aim: Compare the QoL in MCI patients with controls without cognitive impairment, and ascertain whether there are differences in the reports of QoL made by the subjects and by their informants. Methods: Two hundred participants were enrolled, divided into MCI patients (n ¼ 50), MCI informants (n ¼ 50), recruited from a memory clinic and a dementia outpatient clinic, and controls (n ¼ 50) and controls informants (n ¼ 50), recruited in a family practice clinic. QoL was assessed with the QoL in Alzheimer disease (QOL-AD) scale. Results: The total scores of the QOL-AD questionnaire were 32.1 � 6.9 for MCI patients self-report, 27.2 � 6.7 for MCI patients in the opinion of their informants, 35.3 � 4.9 for controls self-report and 35.6 � 4.9 for controls in the opinion of their informants. MCI patients had lower QOL-AD scores than controls. The QoL reported by patients with MCI was more favorable than the opinion of their informants. Conclusion: The QoL is affected at early stages of cognitive decline. The QoL reported by patients with MCI is better than the opinion of their informants, similarly to what is known in Alzheimer's disease patients. QoL appears to be an important domain to be evaluated in aging studies.

Criteria for amnestic MCI rely on the use of delayed recall tasks to establish the presence of memory impairment. This study applied the California Verbal Learning Test to detail memory performance in MCI patients (n = 70), as compared to control subjects (n = 92) and AD patients (n = 21). Learning across the 5 trials was different among the 3 groups. Learning strategy was also different, the MCI group showing less semantic clustering than the control group. However, both MCI patients and controls could benefit from semantic cueing. This study showed that beyond consolidation deficits, MCI patients have marked difficulties in acquisition and recall strategies.

The relationship between memory complaints and objective memory performance remains poorly understood, particularly in young and middle aged people. We studied the relationship between reports of memory complaints and objective memory performance, and the possibility of differentiating good and poor reporters across the lifespan based on concordance between reported abilities and objectively assessed performance. This cross-sectional study enrolled 292 healthy individuals, aged 18 to 87 years, able to perform common activities of daily living and without neurological or psychiatric conditions or systemic diseases likely to interfere with cognition. No correlation between memory complaints, as assessed by the Subjective Memory Complaints scale (SMC) score and the objective memory performance, evaluated by the long-delay free recall (LDFR) score of the California Verbal Learning Test (CVLT), was found, even when grouping the participants by decade. The SMC score was influenced by the presence of depressive symptoms. Participants who were more educated, female and younger tended to have a higher CVLT-LDFR score. Younger subjects were more likely to have good memory performance and report few memory complaints than older subjects. In conclusion, there are differences in the reliability of memory reporting across the lifespan, younger subjects being more likely to correctly report good memory than older subjects.