Machine Learning-Based Decision Support System for Optimal Treatment of Acute Inflammation Response with Specific Patient Conditions

Abstract

This chapter proposes reinforcement learning-based optimal treatment of acute inflammatory response with drug dosage regulation, where the mathematical model of inflammation response is a well-known universal model in the literature. In the treatments, external disturbance and ineffective dosage cases have been considered to strengthen the immune system to prevent possible damage by considering the septic and aseptic dynamics of the inflammation response. In addition, particular reward value functions are proposed for the fast and smooth convergence of stabilization and constraining the inflammation states. From the computational results, value approximation models have been found to be much faster than the tabular reinforcement models, even though septic dynamics are stabilized. The application results are compared in terms of root mean-squared performances of drug dosages and inflammation state, and maximum value of the states, respectively. As a result of the testing phase, the treatment results on the random patients are found as promising for the applications in future not only for acute inflammatory response but also for inflammation-based diseases. © 2024 selection and editorial matter, Bhanu Chander, Koppala Guravaiah, B. Anoop, and G. Kumaravelan; individual chapters, the contributors.