Optimizing Personalized Cancer Treatment Plans using Deep Reinforcement Learning in IoT-Enabled Healthcare Systems

Personalized cancer therapy is vital to patient outcomes, yet traditional methods are often not adaptive to the individual responses of the patient. The IoT-based deep reinforcement learning framework proposed in this study will optimize personalized cancer treatment plans through continuous learning using patient data. Unlike traditional rule-based or AI static models, the method proposed adjusts treatment approaches in real time according to the patient's current health. The framework consists of integrating patientcentric data from sensors, electronic health records, and medical imaging using various advanced DRL algorithms— namely, Deep Q-Networks, Policy Gradient methods, Advantage Actor-Critic, and Proximal Policy Optimization. PPO performed the best with 91.3% accuracy, an 87.9% survival rate, and 89.7% treatment adherence, thus ranking at the top among the works in treatment decision-making. Additional experiments showed a marked shrinkage in tumor size as well as fewer side effects, thereby establishing the superiority of DRL for optimizing treatment. Thus, the function of open adaptive data-driven decision-making in the proposed system ultimately will ensure the implementation of personalized treatment modalities. In essence, this study lays a solid pathway to demonstrate the feasibility of AI-driven personalized cancer therapy within an IoT-enabled healthcare system: optimized treatment decisions can be made in real time from patient information and continuous learning.