Computer-assisted decision making in medicine.

This article reviews the strengths and limitations of five major paradigms of medical computer-assisted decision making (CADM): (1) clinical algorithms, (2) statistical analysis of collections of patient data, (3) mathematical models of physical processes, (4) decision analysis, and (5) symbolic reasoning or artificial intelligence (AI). No one technique is best for all applications, and there is recent promising work which combines two or more established techniques. We emphasize both the inherent power of symbolic reasoning and the promise of artificial intelligence and the other techniques to complement each other.

PUFF: an expert system for interpretation of pulmonary function data.

The application of artificial intelligence techniques to real-world problems has produced promising research results, but seldom has a system become a useful tool in its domain of expertise. Notable exceptions are the DENDRAL (1) and MOLGEN (2) systems. This paper describes PUFF, a program that interprets lung function test data and has become a working tool in the pulmonary physiology lab of a large hospital. Elements of the problem that paved the way for its success are examined, as are significant limitations of the solution that warrant further study.

Managing the data from respiratory measurements.

Clinical decisionmaking depends upon properly interpreting the significance of physiological and other clinical data. Our experience, summarized in six case studies, suggests that no one variable is sufficient for making clinical decisions. Rather, different parameters are relevant in different situations. This article summarizes two techniques for improving the effectiveness of clinical decisionmaking in the ICU using quantitative physiological monitoring data. First, mathematical modeling has been used for measuring the volume of gas in the lungs of patients receiving mechanical ventilation. The technique analyzes the transient response to oxygen change; thus it is suitable for routine use in the ICU. Second, symbolic processing has been used for interpreting the clinical significance of measured data. This symbolic processing is used for recognizing artifact in measured data, determining expected physiological meaning of measured data in different clinical situations, identifying physiological status, and identifying therapy that may be appropriate for meeting therapeutic goals or correcting physiological problems in patients in the intensive care unit.

Recent experience with a respiratory monitoring system in intensive care.

Computer based instrumentation for continuous monitoring of airway flow, pressure, O2 and CO2 concentration offers an improved noninvasive management technique for patients on mechanical ventilators. Computation of these basic signals provides routinely the following measurements: respiratory rate, tidal volume in and out, minute ventilation, positive end-expiratory pressure, mean airway pressure, inspiration-expiration ra measurements, except for O2 consumption and partially for tco2 production. The system works as a monitor of the respirator (detection of malfunction) and as a monitor of the lung function of the patient. It is particularly useful when adjusting the respirator and at time of weaning a patient from the respirator. These maneuvers can be made more safely because they are based on objective measurements and followed by immediate new sets of data. Defining the optimal values of tidal volume and positive end-expiratory pressure has been simplified by the use of pressure-volume plots. A "fighting", is now used as a measure of the severity of "fighting", that is of the effort of the patient to breathe spontaneously while being ventilated. It can detect fighting before it is diagnosed clinically and so can provide a warning that significant physiological changes will occur unless the fighting is controlled. New information about the distribution of ventilation-perfusion ratio can be derived from the expired concentration curve for CO2. Quantitative measurement of the distribution of ventilation shows a very close correlation with clinical events and can be carried out automatically during the normal routine of care of the patient. These on-line quantitative measurements, with the immediate reporting of results, appear to make a positive contribution to patient care.