Performance Comparison of Machine Learning Techniques in Identifying Dementia from Open Access Clinical Datasets

Identified mainly by memory loss and social inability, dementia may result from several different diseases. In the world with ever growing elderly population, the problem of dementia is rising. Despite being one of the prevalent mental health conditions in the community, it is not timely identified, reported and even understood completely. With the massive improvement in the computational power, researchers have developed machine learning (ML) techniques to diagnose and detect neurodegenerative diseases. This current work reports a comparative study of performance of several ML techniques, including support vector machine, logistic regression, artificial neural network, Naive Bayes, decision tree, random forest and K-nearest neighbor, when they are employed in identifying dementia from clinical datasets. It has been found that support vector machine and random forest perform better on datasets coming from open access repositories such as open access series of imaging studies, Alzheimer’s disease neuroimaging initiative and dementia bank datasets.