The primary objective of our project was to predict binary-coded CDR scores in both cross-sectional and longitudinal contexts. Initially, we conducted a joint analysis of OASIS-I and OASIS-II, binarised CDR, and preprocessed input features. Subsequently, we selected and optimised a logistic regression classifier for cross-sectional prediction, achieving balanced accuracies of 80.7% and 78.6% with and without MMSE as a feature, respectively. Finally, we designed and trained an LSTM model that predicted the CDR outcome longitudinally with a balanced accuracy of 80%.