E. Artioli

Variable selection for the classification of postoperative cardiac patients

A set of 13 extensively used hemodynamic, ventilatory and gas analysis variables are measured (on-line or off-line) on 200 patients in an intensive care unit (ICU) during the 6 h immediately following cardiac surgery. Since the existence of two well-separated classes of low- and high-risk patients has been previously shown the divergence criterion is then used to identify those variables which, at three equidistant observation times, possess the greatest separation power. Such variables always include the cardiac index (CI), representative of cardiac performance, and two indices related to respiratory efficiency and metabolic rate, i.e. the carbon dioxide production index (VCO2I) and the arterio-venous oxygen difference (avO2D). The Fisher linear classifier, utilizing these three features, is then tested by using the rotation method. The results show good performance of the linear classifier, which exhibits a probability of correct recognition always greater than 87%, thus suggesting the possibility of obtaining interesting improvements by means of more sophisticated classifiers.

Extraction of discriminant features in post-cardiosurgical intensive care units

A linear transformation, based on the Karhunen-Loève expansion, is applied to 13 physiological variables, measured in 200 surgical patients, in order to extract a limited number of features well representative of the differences between normal and high-risk classes of subjects. This transformation may be considered as a mapping from the primitive 13-dimensional space to a lower dimensional one, without severely reducing class separability. The efficacy of both transformed and primitive variables in the separation of normal and high-risk subjects is compared using the error probability, i.e. the probability that a patient is assigned to the wrong class. In particular, its upper bound is evaluated through the Kullback divergence and its estimate is computed, from the available samples, by applying a quadratic classifier. The results obtained show that only two transformed variables are able to present a divergence better than the most effective set of eight primitive variables. In agreement with the divergence criterion, the classifier provides a recognition error lower than 5% and greater than 13% when using the two best transformed and the two best primitive variables, respectively. Even though the new variables do not have a direct physiological meaning, this limitation has been partially overcome by calculating the correlation matrix between transformed and primitive variables. The results presented show that the first two transformed variables are strongly related to the most discriminant primitive ones (i.e. cardiac index, oxygen delivery and arterio-venous oxygen difference). In conclusion, the transformation of variables proposed appears to be extremely attractive for practical applications, since it allows recognition systems to be designed which exhibit both high performance and great simplicity.

An expert system based on causal knowledge: validation on post-cardiosurgical patients.

A new expert system for the analysis of post-cardiosurgical patients in Intensive Care Units is described, and a preliminary validation performed. The inference engine employs a hybrid reasoning method which integrates quantitative and qualitative simulation techniques in an original manner. The long-term knowledge consists of a causal network which reproduces the main relationships between physiological quantities involved in the course after cardiac surgery. Emphasis has been given to respiratory and metabolic, as well as cardiovascular quantities both in the systemic and pulmonary circulations. Preliminary system validation has been performed on a set of 40 cardiosurgical patients, previously classified either at normal-risk (17 patients) or at high-risk (23 patients) by means of statistical classification techniques. In most cases, predictions of the expert system substantially agree with those provided by the more traditional statistical method. The system, however, is also able to furnish detailed explanations on the possible physiological causes responsible for the patient status. In particular, simulation results indicate that a reduction in the cardiac index (19 cases) and an increase in the oxygen utilization coefficient (19 cases) are the most critical alterations in the high-risk patients. The system imputes the reduced cardiac index to a rise in total systemic resistance (15 high-risk patients), a decrease in cardiac strength (2 high-risk patients) or an insufficient filling volume of the systemic circulation (4 high-risk patients). Furthermore, in 6 high-risk patients the depressed cardiac outflow occurs with a reduction in the arterial oxygen content, mainly imputable to an insufficiency of blood hemoglobin content. Finally, two examples of the complete expert system explanatory capabilities are shown with reference to a pair of high-risk patients and discussed.

A four-parameter linear model for analysing cardiorespiratory data in post-operative cardiac patients

This paper investigates the possibility of characterizing the differences between normal- and high-risk postoperative cardiac patients on the basis of four parameters related to a simple linear model of cardiorespiratory performances. The model comprises three subsystems representing cardiac, vascular and respiratory functions, respectively. These parameters, determined from physiological variables measured in the Intensive Care Unit, seem useful for clinical evaluation of patient status. In fact, their values quantify the improved cardiovascular and respiratory response that normal-risk patients exhibit to increasing metabolic needs after hypothermic treatment, with less utilization of blood oxygen reserve. In addition, a set of three parameters derived from the proposed four allows a prediction of patient class membership with an error lower than 7% when used with a Bayes quadratic classifier.

Classification of postoperative cardiac patients by means of simple neural networks

The aim of this work is to compare the performance of simple neural networks with that of statistical Gaussian classifiers in the discrimination of normal- and high-risk post-operative cardiac patients. The considered data include 13 cardio-respiratory quantities taken from 158 patients. Feature extraction techniques were applied in order to extract the 3 variables out of 13 more suitable for classification. The results indicate that there are combinations of three original variables which allow better discrimination between normal and high-risk patients by neural networks than by a Gaussian classifier. For instance, using the cardiac index, the left ventricle stroke work index and oxygen pressure in mixed venous blood, a linear perceptron can discriminate patients with an error as low as 6.9%. Application of the Karhunen-Loeve transform allows the classification error to be reduced to 5.1% with both neural and statistical classifiers if a non-linear discriminant function is used.

Simple cardiorespiratory model for interpreting post-operative patient data

This paper investigates the possibility of characterizing the differences between normal- and high-risk post-operative cardiac patients on the basis of 3 parameters related to a simple model of cardiorespiratory performances. These parameters, determined from physiological variables measured in Intensive Care Unit (ICU), seem useful for clinical interpretation of patient status. In fact, their values show that normal-risk patients exhibit a better cardiac response to metabolic needs and less utilization of blood oxygen reserve. In addition, they allow effective prediction of patient class membership.

Qualitative simulation of physiological dynamical models involving second-order derivatives

Qualitative simulation of dynamical models is a promising new subject in the field of artificial intelligence, especially suitable for the analysis of clinical and physiological problems. The major limitation of this method, however, consists in the excessive number of alternative solutions arising when systems of order higher than one are simulated. In particular, many solutions produced in the course of qualitative reasoning are inconsistent, i.e. they have no real physical significance. The present study analyses how a second-order dynamical model can be qualitatively simulated, avoiding the occurrence of inconsistent solutions. This Is made possible by including additional qualitative constraints in the model, i.e., constraints which do not merely arise from causal links between quantities, but depend on the peculiar nature of mathematical equations. In particular, we suggest the use of some constraints on the second time derivatives of state variables, evaluated at those instants when the first time derivatives become zero. A simple example, concerning qualitative simulation of a second-order compartmental model is presented and discussed.

A new expert system for evaluating cardiosurgical patients in intensive care units: preliminary validation

A new expert system for the analysis of cardiosurgical patients in Intensive Care Units (ICUs) is described and preliminary validations performed. The system employs a hybrid inference engine. Which integrates both quantitative and qualitative reasoning in an original way. Causal knowledge of the main relationships among hemodynamic, metabolic and respiratory quantities is used to make predictions about the patient status starting from the value of a few quantities monitored in ICUs. Examples concerning six real patients are presented and discussed. In all cases, results of the expert systems substantially agree with those previously obtained using more traditional statistical classification techniques. As an essential improvement, the expert system also provides information on deep physiological quantities and causal relationships which may affect the patient status.

Sensitivity analysis of a simple cardiorespiratory model in postoperative cardiac patients

This paper investigates the influence of four parameters of a simple cardiorespiratory model on relevant physiological variables, such as cardiac output, CO, and oxygen utilization coefficient, CU. In particular, analysis of the sensitivity of these variables to changes on cardiac contractility, CC, blood volume, BV, hemoglobin, Hb, and dead space volume, DS, is described. With this analysis, information is obtained about which parameters are more critical and require particular attention in evaluating the cardiopulmonary performance of postoperative cardiac patients.

Comparative evaluation of two hypothermic treatments in cardiac surgery: a statistical decision theory approach

Blood cardioplegia can be administered both in anterograde and retrograde fashions. In an effort to better define the clinical performance of the combined anterograde/retrograde cardioplegia, this work compares the results yielded by this method with those obtained with the classic asanguineous crystalloid cardioplegia by analyzing their effects on a set of physiological variables monitored in patients. A set of 96 adult patients monitored in the intensive care unit (ICU) after hypothermic cardiopulmonary bypass for coronary or valvular disease was studied. In 55 patients the crystalloid cardioplegia technique was used as hypothermic treatment, while in the other patients the antero/retrograde blood cardioplegic technique was adopted. The statistical analysis developed can be divided into two phases. The first analyzes the separability of the two classes and their evolution in time. The second phase deals with the problem of identifying the single variable and/or the variable subset with the greatest separation power. The results confirm the role of the antero/retrograde blood cardioplegia in myocardium protection.<<ETX>>