Development of a knowledge-base for automatic monitoring of renal functionof intensive care patients over time

Renal dysfunction is a major problem in the management of critically ill pa tients. Monitoring of renal parameters over time is a prerequisite for dete ction of any significant deterioration of kidney function. Thus, we develop ed a knowledge-base for the dynamic monitoring of renal function of critica lly ill patients. A database with renal parameters of 750 intensive care pa tients was analyzed for distribution of parameters within predefined interv als of the creatinine clearance. Additionally, a subgroup of 11 patients wi th (quite) normal renal function over 11 days was selected and the daily va riability of renal parameters was analyzed. An interdisciplinary expert tea m selected a set of nine clinically relevant renal parameters and formulate d, on the basis of the data analysis and the parameter set, eight definitio ns of renal function, which represent four levels of renal performance. The se definitions were arranged into an hierarchical structure, considering on ly clinically relevant changes of renal function. A change from one functio nal state to another inside of 2 days indicates a relevant alteration of re nal function. Monitoring of time courses can additionally be performed by s tatistical analysis of the daily variability of parameters and comparison w ith their 'normal' variability. Moreover, rules were established for the pl ausibility check of results and interpretations of single parameters and pa rameter sets formulated. (C) 2000 Elsevier Science Ireland Ltd. All rights reserved.