Alzheimer’s disease (AD) is a degenerative neurological disease that is the most common cause of dementia. It is also the fifth-greatest reason for death in adults aged 65 and over. However, there is no accurate way of diagnosing neurological Alzheimer’s disorders in medical research. Blood gene expression analysis offers a realistic option for identifying those at risk of AD. Blood gene expression patterns have previously proved beneficial in diagnosing several brain disorders, despite the blood-brain barrier’s restricted permeability. The most extensively used statistical machine learning and deep learning algorithms are data-driven and do not address data uncertainty. Belief Rule-Based Expert System (BRBES) is an approach that can identify various forms of uncertainty in data and reason using evidential reasoning. No previous research studies have examined BRBES’ performance in diagnosing AD. As a result, this study aims to identify how effective BRBES is at diagnosing Alzheimer’s disease from blood gene expression data. We used a gradient-free technique to optimize the BRBES because prior research had shown the limits of gradient-based optimization. We have also attempted to address the class imbalance problem using BRBES’ consequent utility parameters. Finally, after 5-fold cross-validation, we compared our model to three classic ML models, finding that our model had a greater specificity than the other three models across all folds. The average specificity of our models for all folds was 32%