Comparative machine learning approach in dementia patient classification using principal component analysis

Dementia is one of the brain diseases that were significantly affecting the global population. Mainly it is exposed to older people with an association of memory loss and thinking ability. Unfortunately, there are no proper medications for dementia prevention. Doctors are suggesting that early prediction of this disease can somehow help the patient by slowdown the dementia progress. Nowadays, many computer scientists were using machine learning (ML) algorithms and data-mining operations in the healthcare environment for predicting and diagnosing diseases. The current study designed to develop an ML model for better classification of patients associated with dementia. For that, we developed a feature extraction method with the involvement of three supervised ML techniques such as support vector machines (SVM), K-nearest neighbor (KNN), and logistic regression (LR). Principal component analysis (PCA) was selected to extract relevant features related to the targeted outcome. Performance measures were assessed with accuracy, precision, recall, and AUC values. The accuracy of SVM, LR, and KNN was found as 0.967, 0.983, and 0.976, respectively. The AUC of LR (0.997) and KNN (0.966) were recorded the highest values. With the highest AUC values, KNN and LR were considered optimal classifiers in dementia prediction.