Approximately 1% of the world population is afflicted with Epilepsy. For many patients, antiepileptic drugs do not fully control seizures. Electrical brain stimulation therapies have been effective in reducing seizure rates in some patients. While current neuromodulation devices provide a benefit to patients, efficacy can be improved by optimizing brain stimulation so that the therapy is tuned on a patient by patient basis. One optimization approach is to target deep brain regions that strongly modulate seizure prone regions. I will present data on the effects of stimulation of two different anatomical regions for seizure control, and establish my experimental platform for testing closed-loop algorithms. There are two general methods to implementing closed-loop algorithms to modulate neural activity: 1) Model-free algorithms that require a learning period to establish an optimal mapping between neural states and best therapeutic parameters, and 2) Model-based algorithms that use forward predictions of the neural system to determine the appropriate stimulation therapy to be administered. In this thesis, I will propose and test two closed-loop control schemes to control the brain activity to prevent epileptogenic activity while reducing stimulation energy. I will also present techniques to remove stimulation artifacts so that neural biomarkers can be measured while simultaneously applying stimulation. The methods I will present could potentially be implemented in next generation electrical brain stimulation hardware for seizure disorders and other neurological diseases.