Dementia is a fast-growing public health problem, with an estimated 47 million people currently living with the condition. By 2030, this total is predicted to reach 75 million. By 2050, it will have tripled were, given the urgent need to address this problem. Alzheimer's disease is characterized by a steady decline in cognitive capacities beginning with a decrease in the brain's capacity to form new memories. Significant attention has been focused on developing therapeutic strategies and drugs to treat Alzheimer's disease, which is the most common form of dementia. In this study, the feature used is the PubChem Fingerprint representing the molecule's structure with a total of 822 data for class 0 and 691 data for class 1. We developed a fingerprint-based artificial neural network (ANN) model to predict Beta-secretase 1 (BACE-1) inhibitors as therapeutic agents for Alzheimer's disease. Three optimization strategies, namely the Bat Algorithm, the Hybrid Bat Algorithm, and the Adaptive Bat Algorithm, were used to optimize the architecture of the ANN. This nature-inspired optimization technique mimics the echolocation behavior of bats. The best model was obtained from ANN optimized using Hybrid Bat Algorithm with the value of accuracy and F1-score are 0.81 and 0.78, respectively.