Predicting the Progression from Clinical Normal and Mild Cognitive Impairment to Alzheimer’s Disease using Socio-Demographic, Non-Clinical, and Genetic Data with Multi-Classification Models

Dementia and cognitive impairment are the leading causes of illness in the elderly. The research objective is to predict whether the subject is Clinical Normal (CN), Mild Cognitive Impairment (MCI), or Dementia, and when the progression from MCI to Dementia occurs within four years by examining socio-demographic, clinical, and genetic factors. The author used Alzheimer’s Disease data, containing socio-demographic, non-clinical assessment, and genetic data. A grid search approach was employed to find the best classification model. Demonstrated that Support Vector Classifier performs significantly better than other classifiers, achieving F1-score of 0.78. The proposed classifier, when combined with genetic factors, improves the overall result of multi-classification, yielding an F1-score of 0.84. The study successfully identified the impact of genetic factors on the progression of dementia. These findings can contribute to lowering the disease burden on individuals, reducing healthcare costs, and aiding in the discovery of dementia-modifying therapies.