NEW DECISION SUPPORTS AND DATABASES FOR DRUG DOSAGE

点击翻译按钮获取中文摘要

Precise drug therapy demands dosage regimens to achieve and maintain selected therapeutic goals as precisely as possible, minimizing the well- known variability in patient response. In the past 3 years we developed practical methods to model population pharmacokinetic behavior for most drugs used singly, determining the entire probability distribution of the parameter values, making no conventional assumptions as to the shape of those distributions, capable of discovering fast and slow metabolizers, for example, when they were not anticipated. This NPEM2 program is operational. It provides useful databases to describe drug behavior. It has not been coupled with our new 'multiple model' (MM) dosage designer for decision support in drug dosage, which specifically minimizes patient variability in response about the selected therapeutic goals. These are breakthroughs. Clinical studies are in progress. We now propose to build on these successes to implement a clinical graphical user interface for the MM dosage designer to make it usable by clinicians in community hospitals, and to extend the application of these methods from 3 compartment linear models to large linear and nonlinear pharmacokinetic and pharmacodynamic (effect) models, to better control toxic side effects, for example, seen in treatment of infections, AIDS, and cancer. This will optimize coordinated therapy with multiple drugs, holding their common toxic effects within stated tolerable limits (platelets not below 35,000, for example). We have also implemented parameters which serve as database for documenting the quality of care each patient receives with these drugs and which each center gives (environmental noise parameters). We will extend these to large linear models as well. Further, we will implement, based on the success of the MM dosage designer, a new 'Active' dosage designer which uses both the dose and the serum data as active partners in the learning process, taking advantage of the fact that we know we will monitor the regimen after it is given. Instead of learning only from past mistakes, this new designer used both the dosage and the anticipated serum data together to play out, in advance, many therapeutic scenarios, to optimize, for the first time, the process of having to learn about the patient while having to treat him/her at the same time. We will also implement these population modeling and control methods into practical software for clinical use in community hospitals, to continue to improve patient care, to optimize patient care and further shorten hospital stay.