SCH: EAGER: Personalizing Drug Delivery through Clinically Relevant Modeling and Control

University Of Louisville Research Foundation Inc, Louisville KY

Investigators

Abstract

The variety and complexity of medications used by physicians for the treatment of diseases are increasing. Difficulties arise in finding the proper dose for drugs that have a narrow therapeutic range, high toxicity, altered effects with disease progression, or those that are affected by genomic factors. Because of these complexities, physicians often rely on trial-and-error-like approaches to determine optimal drug doses for individual subjects. Unfortunately, these approaches may lead to over- and under- dosing, resulting in subtherapeutic effect or increased risk of adverse events. Advances in engineering and computational intelligence may help to alleviate much of the uncertainty when dosing these agents. To address the complexities in individualized drug dose, this EAGER research aims to investigate a new methodology for the development of individualized patient models and a novel control approach for personalized drug dosing. In the short term, this effort will impact the treatment of anemia, a common chronic condition in hemodialysis patients. The successful completion of the proposed research, however, has the potential to result in a new method for personalized drug dosing that can be applicable to many other chronically administered drugs that require some type of monitoring. On the educational front, plans include training graduate students and involving undergraduate students with research. Several existing programs such as INSPIRE and the High School Outreach Program will be used to reach out to high school students. Outcomes of the project will be made available online. The objective of this EAGER project is to investigate a novel approach to individualized drug dosing using a new modern control method based on Radial Basis Functions (RBF) and robust patient-specific models. The project has the following three specific aims: 1) The theoretical principles of semi-blind robust system identification are examined to produce individual patient models from a limited number of noisy patient-specific clinical data. In contrast to classical identification techniques, robust system identification takes into account system uncertainties, unmodeled dynamics and model complexity, i.e., there is no assumption on the model order, uncertainties and noise affecting data. 2) Based on the these predictive individualized models, the development of a patient-specific dosing approach using a novel RBF-Galerkin controller is proposed to generate optimal dosing sequence and expected response adapting to temporal changes in patients' dose-response characteristics. 3) The new method will be evaluated in silico using a test case, derived from previously collected clinical data: Erythropoietin (EPO) dosing for anemia management. This is a clinically relevant test-bed as the administration of drug EPO is intensively monitored, has narrow therapeutic range and can have serious adverse effects. Following successful in silico evaluation of the proposed method, a validation study will be performed at the University of Louisville controlled hemodialysis unit. As a benchmark we will use existing state-of-the art approaches to anemia management as well. Successful application of the proposed technique to this agent will provide proof-of-concept that the proposed method can be applicable in clinics. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.