Early detection and prognosis of dementia is a critical global health challenge given the rapid aging of populations. Despite the increase in dementia, there is a significant gap in research on early detection and prognosis of dementia, particularly utilizing artificial intelligence (AI) and machine learning (ML) techniques for more accurate and timely diagnosis. This study explores the efficacy of multiple ML algorithms in predicting the onset of dementia using the OASIS dataset, which contains extensive MRI data of individuals aged 60–96 years. Eight state-of-the-art predictive AI/ML algorithms are applied to evaluate dementia classification including support vector machine (SVM), K-nearest neighbors (KNN), decision trees (DT), gradient boosting machine (GBM), random forest (RF), Naïve Bayes (NB), logistics regression (LR), and extra trees (ET). Among the tested models, SVM achieved the overall best performance, with 92% accuracy, 0.92 specificity, and an F1-score of 0.91, highlighting its potential for early dementia classification. Key features such as the mini-mental state examination (MMSE) scores and clinical dementia rating (CDR) are used to train the models. Feature selection was achieved using LASSO and Fisher’s exact test. The results indicate that the models achieved high accuracy along with other performance metrics, which highlights the potential of ML techniques in enhancing diagnostic processes and informing early intervention strategies for dementia, potentially improving patient outcomes.