Enhancing MCI Detection with a Hybrid Machine Learning Approach

Abstract

Mild Cognitive Impairment (MCI) is a prodromal stage of dementia, is often native with very high risk of evolving into Alzheimer's disease. Early detection and accurate classification of MCI can significantly aid in timely intervention and personalized treatment planning. In the study conducted, we put forward a hybrid machine learning approach to enhancing MCI detection using a combination of feature engineering, feature selection, and ensemble learning algorithm. by using standalone Recurrent Neural Networks (RNN) as well as Convolutional Neural Networks (CNN). The proposed method leverages the temporal dependencies captured by RNN and the spatial information extracted by CNN to increase the robustness and accuracy of MCI classification. We used a comprehensive dataset ADNI consisting of neuroimaging and clinical data from a large cohort of subjects, including MCI sufferers and healthy controls. The neuroimaging data encompassed structural MRI scans, while the clinical data encompassed various cognitive assessments. Neuroimaging data is preprocessed to extract relevant features and combine them with the clinical data to create a unified input representation for the hybrid model.