Artificial Intelligence Based Videogame To Treat Patients With Schizophrenia

Machine learning algorithms can be broadly classified into supervised, unsupervised and reinforcement learning models. The well-suited technique for video game using machine learning is reinforcement learning. Reinforcement learning algorithm is about taking suitable action to maximize reward in any particular situation. Here there is no label, the agent decides what to do to perform given any task. Training process is not required. Patients with schizophrenia tend to experience an array of symptoms that affect their mental health. They do not co-operate with the treatment and loose hope quite often and very fast. Q learning is a reinforcement-based algorithm. The main advantage is that it is model free algorithm with a Q table that is updated using the bellman equation. This updating is done in order to maximize the rewards obtained. It is also dependent on various hyper parameters that are used to control the behavior of the algorithm. This maximum reward is then used to guide the agent to choose the best possible action at any given point of time. As the agent always chooses the best possible action based on the maximum reward, dynamicity is introduced in the game being played and also the winning probabilities are increased. This supports in treatment of patients as they would always win the game.