Estimulação cerebral profunda adaptativa baseada em aprendizado por reforço: um modelo computacional de neurônios de Izhikevich para simulação do tratamento dos tremores da doença de Parkinson

Abstract

The dissertation delves into the development of an adaptive computational model for the simu- lation of treating tremors associated with Parkinson’s disease, using Deep Brain Stimulation technique in combination with reinforcement learning strategies and neuron modeling through the Izhikevich model. The primary goal is to optimize therapeutic outcomes potentially transfor- ming the treatment paradigm for Parkinson’s patients. This study stands out for its personalized approach to treatment, proposing a methodology that dynamically adjusts stimulation parameters in response to the patient’s varying conditions. The findings are analyzed in terms of symptom minimization efficiency and energy expenditure. Furthermore, a comparison is made with com- mercially established brain stimulation standards, where the model presented here is potentially more energy-efficient when coupled with signal treatment techniques, showing an 80% rate of energy released in comparison to the commercial model.