Marion Verduijn

Prognostic Bayesian networks

Prognostic models are tools to predict the future outcome of disease and disease treatment, one of the fundamental tasks in clinical medicine. This article presents the prognostic Bayesian network (PBN) as a new type of prognostic model that builds on the Bayesian network methodology, and implements a dynamic, process-oriented view on prognosis. A PBN describes the mutual relationships between variables that come into play during subsequent stages of a care process and a clinical outcome. A dedicated procedure for inducing these networks from clinical data is presented. In this procedure, the network is composed of a collection of local supervised learning models that are recursively learned from the data. The procedure optimizes performance of the networks primary task, outcome prediction, and handles the fact that patients may drop out of the process in earlier stages. Furthermore, the article describes how PBNs can be applied to solve a number of information problems that are related to medical prognosis.

Temporal abstraction for feature extraction: A comparative case study in prediction from intensive care monitoring data

OBJECTIVES To compare two temporal abstraction procedures for the extraction of meta features from monitoring data. Feature extraction prior to predictive modeling is a common strategy in prediction from temporal data. A fundamental dilemma in this strategy, however, is the extent to which the extraction should be guided by domain knowledge, and to which extent it should be guided by the available data. The two temporal abstraction procedures compared in this case study differ in this respect. METHODS AND MATERIAL The first temporal abstraction procedure derives symbolic descriptions from the data that are predefined using existing concepts from the medical language. In the second procedure, a large space of numerical meta features is searched through to discover relevant features from the data. These procedures were applied to a prediction problem from intensive care monitoring data. The predictive value of the resulting meta features were compared, and based on each type of features, a class probability tree model was developed. RESULTS The numerical meta features extracted by the second procedure were found to be more informative than the symbolic meta features of the first procedure in the case study, and a superior predictive performance was observed for the associated tree model. CONCLUSION The findings indicate that for prediction from monitoring data, induction of numerical meta features from data is preferable to extraction of symbolic meta features using existing clinical concepts.

Prognostic Bayesian networks: II: An application in the domain of cardiac surgery

A prognostic Bayesian network (PBN) is new type of prognostic model that implements a dynamic, process-oriented view on prognosis. In a companion article, the rationale of the PBN is described, and a dedicated learning procedure is presented. This article presents an application here of in the domain of cardiac surgery. A PBN is induced from clinical data of cardiac surgical patients using the proposed learning procedure; hospital mortality is used as outcome variable. The predictive performance of the PBN is evaluated on an independent test set, and results were compared to the performance of a network that was induced using a standard algorithm where candidate networks are selected using the minimal description length principle. The PBN is embedded in the prognostic system ProCarSur; a prototype of this system is presented. This application shows PBNs as a useful prognostic tool in medical processes. In addition, the article shows the added value of the PBN learning procedure.

Individual and Joint Expert Judgments as Reference Standards in Artifact Detection

OBJECTIVE To investigate the agreement among clinical experts in their judgments of monitoring data with respect to artifacts, and to examine the effect of reference standards that consist of individual and joint expert judgments on the performance of artifact filters. DESIGN Individual judgments of four physicians, a majority vote judgment, and a consensus judgment were obtained for 30 time series of three monitoring variables: mean arterial blood pressure (ABPm), central venous pressure (CVP), and heart rate (HR). The individual and joint judgments were used to tune three existing automated filtering methods and to evaluate the performance of the resulting filters. MEASUREMENTS The interrater agreement was calculated in terms of positive specific agreement (PSA). The performance of the artifact filters was quantified in terms of sensitivity and positive predictive value (PPV). RESULTS PSA values between 0.33 and 0.85 were observed among clinical experts in their selection of artifacts, with relatively high values for CVP data. Artifact filters developed using judgments of individual experts were found to moderately generalize to new time series and other experts; sensitivity values ranged from 0.40 to 0.60 for ABPm and HR filters (PPV: 0.57-0.84), and from 0.63 to 0.80 for CVP filters (PPV: 0.71-0.86). A higher performance value for the filters was found for the three variable types when joint judgments were used for tuning the filtering methods. CONCLUSION Given the disagreement among experts in their individual judgment of monitoring data with respect to artifacts, the use of joint reference standards obtained from multiple experts is recommended for development of automatic artifact filters.

Modeling length of stay as an optimized two-class prediction problem.

OBJECTIVES To develop a predictive model for the outcome length of stay at the Intensive Care Unit (ICU LOS), including the choice of an optimal dichotomization threshold for this outcome. Reduction of prediction problems of this type of outcome to a two-class problem is a common strategy to identify high-risk patients. METHODS Threshold selection and model development are performed simultaneously. From the range of possible threshold values, the value is chosen for which the corresponding predictive model has maximal precision based on the data. To compare the precision of models for different dichotomizations of the outcome, the MALOR performance statistic is introduced. This statistic is insensitive to the prevalence of positive cases in a two-class prediction problem. RESULTS The procedure is applied to data from cardiac surgery patients to dichotomize the outcome ICU LOS. The class probability tree method is used to develop predictive models. Within our data, the best model precision is found at the threshold of seven days. CONCLUSIONS The presented method extends existing procedures for predictive modeling with optimization of the outcome definition for predictive purposes. The method can be applied to all prediction problems where the outcome variable needs to be dichotomized, and is insensitive to changes in the prevalence of positive cases with different dichotomization thresholds.