Model Predictive Control of an HBV Model Based on a Nonlinear Kalman Filter

초록

In this talk, we derive a guideline for efficient drug treatment strategies for hepatitis B virus (HBV) infection. We introduce and analyze a mathematical model that describes the HBV infection during antiviral therapy. The reproduction number R0 is determined. The local/global stability of virus-free steady state is investigated. We formulate a control problem which minimizes the viral load as well as treatment costs. In order to reflect the status of patients not only at the initial time but also at the follow-up visits, we consider the model predictive control based on ensemble Kalman filter and differential evolution. The ensemble Kalman filter is employed to estimate full information of the state from incomplete observation data [1]. We derive piecewise constant drug schedule applying techniques of differential evolution algorithm [2]. Numerical simulations are performed using various weights in the objective functional to suggest optimal treatment strategies in different situations.