Optimal Administration of Anticoagulant Drugs(2nd Report)

Abstract

We previously reported on an optimal administration method of anticoagulantdrugs solved mathematically for an average patient. In this report, we suggest anoptimal administration method to reach and maintain the desired prothrombin levelsfor each individual patient with unknown parameters, which is the case in clinical applications. For the first two days, the standard administration criteria should be applied in principle since the data is insufficient to identify the three parameters of thepatient model. From the third day, parameter identification is theoretically possible, but the least square method was not successfully applicable, as it was found to selectunrealistic values of no use. We had to choose, as a most direct method, a best-fitmodel from models with wide varieties of parameter sets by comparing the model output and the clinical data. Simulation using patient models with two kinds of noiselevels has successfully explained this situation of parameter identification and thepossibility of regulating prothrombin levels by our identification and administrationmethod.