Evaluation of different stages of dementia employing neuropsychological and machine learning techniques

Data mining in medicine is an emerging field of high importance for providing prognosis and a deeper understanding of the classification of disease. The proposed system concentrates on the two different cognitive tests for the diagnosis of different stages of dementia. Dementia is considered as the fourth most common disorder among the elderly. Early detection of dementia and correct staging of the severity of dementia is very much required for the further treatment. Machine Learning (ML) and Neural Networks (NN) approaches in data mining finds its applications in various fields, in the present study the applications of ML as well as NN methods used for classifying dementia states to improve accuracy over current dementia screening tools. The Mini Mental State Examination (MMSE) and Functional Activities Questionnaire (FAQ) are recommended tools for screening of patients by the Agency for Health Care Policy Research (AHCPR).Both ML and NN methods improve the classification accuracy for MMSE and FAQ. When compared to the accuracy obtained from AHCPR. By combining both the tests along with ML and NN it is observed that the accuracy can be optimized. ©2009 IEEE