Differential Diagnosis of Dementia: A Knowledge Discovery and Data Mining (KDD) Approach

We are applying Knowledge Discovery and Data Mining (KDD) methods in conjunction with Electronic Medical Records (EMRs) of normally aging and demented subjects to automate the screening and differential diagnosis of Alzheimer's Disease (AD), Vascular Dementia (VD) and other causes. Having successfully developed dementia screening tools with KDD methods, this report describes the extension of these techniques to the harder task of differential diagnosis. We show that the domain of neuropsychologic test performance helps diagnose AD, but not VD, and that additional domains are needed for accurate diagnosis. An additional benefit of KDD methods applied to EMRs includes detecting subtle data entry errors.