| Summary: | 碩士 === 銘傳大學 === 應用統計與資料科學學系碩士班 === 107 === Levey-Jennings control chart for quality control in clinical medical laboratories has been used in decades. In the process of quality control of blood routine test, quality control reagents needed to be replaced monthly for avoiding decay in regular. However, a fault may occur when the concentration or dosage of new reagent differs from the original one because of coming from different batches. It will result in the control limits are not strict enough if these deviations were noticed only when they are out of original setting control range.
In this study, a blood quality control data in 2017 was collected from a medical examination laboratory in hospital located in Taoyuan. Quality control data including white blood cells (WBC), red blood cells (RBC), hemoglobin (Hb) and platelet (PLT) recorded by a laboratory blood analyzer was arranged and analyzed. Using the empirical Bayesian method, we estimated the variation of concentrations of last and current batches to establish a novel control chart with an adjusted upper and lower limit for current batch, and compared the results with the traditional Shewhart method. The average run length (ARL) and sensitivity of the empirical Bayesian method were explored as well.
The study found that ARL showed a qualified capability for the four blood routine tests while using the empirical Bayesian method. Compare to the Levey-Jennings control chart, the novel control chart can show an alert earlier while a deviation occurring and show a fake alert latter when there is no deviation. Parallel tests showed that the longer the time go through, the better the capability can get. We concluded that the empirical Bayesian method can be applied effectively and can improve the capability of quality control in blood examination.
|
|---|
