Iteration and interaction in computer data bank analysis: a case study in the physiologic classification and assessment of the critically ill.

Abstract Present methodology for arriving at data-inferred categories covers a broad area of loosely related techniques, objectives, and concepts. The iterative and interactive nature of the analytic process and the role of the computer in facilitating this analysis are described in a case study of a real problem—the assessment of the physiologic status of critically ill patients. Three stages in the evolution of a severe infectious process producing a shock state in these patients were identified based on quantitative physiologic measurements. These stages were interpreted in the larger frame of reference provided by the clinical record and medical experience in a manner which has important clinical implications.