The detection of sudden change in renal function by time-series analysis, including the use of the Kalman Filter - a new statistical approach with potential applications for chronobiologists
During the development of renal failure body fluid constituents remain in a dynamic state, with fluctuations in the composition of body fluids resulting from food intake, circadian variations in renal function and other factors. There are progressive rises in some body constituents and these are used by physicians to measure the progress of renal disorders, and to evaluate responses to therapy. Pathological change usually progresses systematically. Sequential measurements can be analysed mathematically to allow predictions of outcome, and to define statistically whether treatment may have altered natural history. Mini and microcomputers make graphical presentations of progress instantly available to the clinician, and allow the use of new statistical methods developed for on-line analysis of data. These detect change in established trends and can be used to alert physicians to events of clinical importance, and to make decisions utilising laboratory data more objective and consistent. They can also be used by those studying rhythmic patterns in the evolution of a disease to detect events prior to analysing these events by one of the conventional methods of time-series analysis used to demonstrate rhythmicity. To achieve the potential of new and old methods it is important to reduce errors in measurement and collection to a minimum, and to incorporate information about systematic biological change, into the statistical analyses. Consideration of the changes in renal function in patients with renal failure and after transplantation provide diagnostically useful information but can also be the basis for studies into the chronobiology of renal disorders. The methods described here have a wide potential use for the analysis of time-related data. © 1982.
