Premium
P4‐085: An exploratory analysis of variables associated with MCI evolution through data mining with novel artificial neural networks
Author(s) -
Montali Arianna,
Grossi Enzo,
Concari Letizia,
Copelli Sandra,
Dieci Francesca,
Messa Giovanni,
Caffarra Paolo
Publication year - 2009
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2009.04.854
Subject(s) - stroop effect , recall , neuropsychology , dementia , psychology , association (psychology) , cognition , audiology , test (biology) , medicine , disease , cognitive psychology , psychiatry , paleontology , psychotherapist , biology
(IQ-CODE). Methods: In BASEL neuropsychological data, using the Consortium to Establish a Registry for Alzheimer’s Disease Neuropsychological Assessment Battery (CERAD-NAB) plus Trail Making Test and phonemic fluency (S-words) were collected from 717 non-demented elderly individuals (45% women, age at baseline1⁄470.967.7; years of education 1⁄4 12.363.0; Mini Mental Status Examination (MMSE) at baseline1⁄4 28.7 61.4) and followed up bi-annually (after 2 years: 717 individuals; after 4 years: 622; after 6 six years: 526). Independent samples of comprehensively diagnosed MCI subjects (N1⁄487; 43% women; age1⁄472.068.1, education 10.6þ2.8; MMSE 1⁄426.061.9) and patients with probable AD (N1⁄4153; 65% women age1⁄477.965.7, education1⁄411.162.6; MMSE1⁄423.063.2) were used to create algorithm families utilizing age-, educationand gender-corrected standard scores of the CERAD-plus-NAB and the IQ-CODE. Results: Systematic variations of selected parameters (eg, cut-off score for impairment or number of test scores below cut-off) revealed different percentages of MCI or dementia. The most appropriate algorithm (ie, MCI1⁄4 IQ-CODE< 3.6 and 2/10 CERAD-NAB standard scores -1.5; AD1⁄4 IQ-CODE 3.6 and 2/10 CERAD-NAB standard scores -1.5) resulted in a correct classification rate between MCI and AD of 60%. Using this algorithm, 109/717, ie, 15.2% of BASEL participants were identified as MCI. Conclusions: A systematic search strategy using algorithm families produced new diagnostic procedures in order to retrospectively diagnose MCI and dementia patients in longitudinal studies. Our preliminary most appropriate algorithm will have to be validated in additional independent samples.