TIME-SERIES ANALYSIS IN INTENSIVE-CARE MEDICINE

Objectives: Time series analysis techniques facilitate statistical ana lysis of variables in the course of time. Continuous monitoring of the critically ill in intensive care offers an especially wide range of a pplications. In an open clinical study time series analysis was applie d to the monitoring of lab variables after liver surgery, and to suppo rt clinical decision making in the treatment of acute respiratory dist ress syndrome. Patients and results: For the analysis of lab variables (blood lactate) in 19 patients after liver resections ARIMA (Auto Reg ressive Integrated Moving Average) models were developed for an estima tion period of at least 14 measurements. Prediction values from these models for the following data points were then compared to the actual lab values. With these models in all cases of hepatic complications pa thological changes in the lab values could be differentiated from rand om variance. In 25 patients with ARDS the effects of therapeutic inter ventions on pulmonary target variables (PVR, Q(S)/Q(T), AaDO(2)) was e stimated with interrupted ARIMA models. The time series before the the rapeutic intervention was compared to changes under intervention using the same model including an intervention regressor. With all therapeu tic interventions clinically relevant therapeutic effects could be sta tistically identified in all patients. Similarly, non-effective therap eutic maneuvers could be detected early, eventually changing therapeut ic strategy. Conclusions: Even on the basis of short time series of in tensive care monitoring variables ARIMA models could be successfully e mployed for the analysis of lab variables and therapeutic intervention s. Nevertheless, due to high demands for manpower and to statistical m ethodological limitations the general use of this methodology in clini cal practice apart from controlled clinical studies cannot be recommen ded today.