High-Accuracy Estimation of the Release of Chemotherapeutics from the Core of Polymeric Micelles

In this paper, the estimation of the release of chemotherapeutics from the core of polymeric micelles is addressed. First, the dynamic model of the system is described. Subsequently, the measurement system and measurement model is discussed. The problem of the inherent noise on the dynamic model and the measurement model is shown. The statistics of these noise sequences are usually assumed to be Gaussian with zero mean and a certain standard deviation. However, these statistics are usually not known a priori. Even if these statistics are known at the beginning of system operation, the statistics might change in time due to the change in the operating conditions or due to aging of the system components or the measurement sensors. Therefore, we start by designing an extended Kalman filter to estimate the drug release. Subsequently, to account for the unknown noise statistics, a multiple model adaptive filter is proposed to estimate the drug release. In this multiple model approach, a number of Kalman filters are utilized, each with a certain hypothesis for the measurement noise statistics, to estimate the state of the system. The estimates of the different Kalman filters are subsequently probabilistically weighted to find the optimal estimate of the state. The performance of the proposed multiple model adaptive filter is compared to the performance of the classical extended Kalman filter.