A Comparative Study on Reinforcement Learning in Disease Prediction on Medical Data

Medical imaging has been used extensively in healthcare in recent years for a variety of purposes, including disease diagnosis, treatment planning, and tracking the course of an illness. These applications entail taking pictures of the afflicted organ using a variety of modalities. Image segmentation and classification is two important process performed in disease diagnostic application. Deep reinforcement learning approach applied in many gaming application. DRL provides the accurate result in many real world applications, so researchers pay attention in DRL in medical data that helps the physicians in treating the patient. This study focus on surveying the various applications such as Image registration, lesion localization, image segmentation, image classification and Landmark detection using deep reinforcement learning in healthcare. In addition, this study investigates on the various DRL algorithms (Qlearning, deep deterministic policy gradient (DDPG), Twin Delayed Deep Deterministic Policy Gradient (TDDDPG), Deep Q Network (DQN) and Soft Actor-Critic (SAC) in Alzheimer's disease prediction. Among the DRL methods the DDPG achieved the highest accuracy in detecting the Alzheimer's disease with 97%.