Abstract
Background: Deep brain stimulation (DBS) is employed to adjust the activity of impaired brain circuits. The variability in clinical trial outcomes for treating Alzheimer's disease with memantine is not yet fully understood. We conducted a randomized in silico study comparing virtual DBS therapies with treatment involving an NMDA antagonist combined with DBS in patients with Alzheimer's disease. Methods: Neural network models representing Alzheimer's disease (AD) patients were randomly assigned to four groups: AD, memantine treatment, DBS, and DBS and memantine. Out of 100 unique neural networks created to model moderate and severe AD with varying hippocampal synaptic loss, 20 were randomly selected to represent AD patients. Virtual treatments-memantine, DBS, and DBS and memantine-were applied, resulting in a total of 80 simulations. Results: The normalized mean number of spikes in the CA1 region among the virtual AD hippocampi treated with memantine, DBS therapy, and DBS and memantine differed significantly (p < 0.0001). The normalized mean number of spikes in the virtual AD hippocampi was 0.33 (95% CI, 0.29-0.36) and was significantly lower compared to the number of spikes in the virtual AD hippocampi treated with memantine, which was 0.53 (95% CI, 0.48-0.59) (p = 0.0162), and in the DBS and memantine group, which was 0.67 (95% CI, 0.57-0.78) (p = 0.0001). Conclusions: Our simulation results indicate the effectiveness of virtual memantine and DBS therapy compared to memantine monotherapy for Alzheimer's disease.