|
Abstract:
|
Hemodialysis is a common dialysis procedure for patients whose kidneys can no longer remove enough waste and fluid from the blood. The standard hemodialysis treatment, with 4 hours each session and 3 sessions each week, is a long, painful and costly process for patients. Particularly, the uniform 4-hour duration completely ignores the receptiveness difference between patients. In this work, we propose to monitor the progress of a hemodialysis treatment through the analysis of Raman spectra of used dialysate. Each Raman spectrum represents the molecular composition of a used dialysate sample. Therefore we consider a functional data modeling of the spectra and develop a testing procedure that can detect significant molecular composition changes from one sample to the other. The theoretical and numerical studies on the properties of the test statistic will be provided.
|
