Catálogo de publicaciones - libros

In this paper we introduce to model states and transitions about different cognitive processes that occur during a clinical reasoning activity. A state of the automaton represents a particular process in a complex patient diagnosis using influence diagrams encoding clinical knowledge about the case. Transitions model switch between diagnosis cognitive processes, such as collecting evidences, formulating hypothesis or explicitly asking for assistance at a given point during the reasoning process. That way, we can efficiently model tutoring feedback hints for clinical reasoning learning that are based not only on the clinical knowledge, but also on the sequencing of the tutoring processes.

There are two types of knowledge – declarative (theoretical) and procedural (practical skills). While the former knowledge may be acquired by reading books, the latter requires long intensive practice. The majority of computer-aided learning systems teach declarative knowledge only. This paper presents basic ideas of building intellectual computer systems for teaching procedural expert knowledge, such as medical diagnostic skills. Two sub-problems are under consideration – the elicitation of experienced physician’s decision rules and the construction of the computer system for teaching these rules. Such systems utilize the principle of implicit learning. The authors present the methodology of practical realization of these ideas in application of teaching the art of acute cardiac infarction diagnosis.

A knowledge management perspective is rarely used to model a process. Using the cognitive perspective on knowledge management in which we start our analysis with events and knowledge (bottom-up) instead of with processes and units (top-down), we propose a new approach for redesigning hospital logistic processes. To increase the care efficiency of multi-disciplinary patients, tailored knowledge in content and type that supports the reorganization of care should be provided. We discuss the advantages of several techniques in providing robust knowledge about the logistic hospital process by employing electronic patient records (EPR’s) and diagnosis treatment combinations (DTC’s).

In this paper, we propose an automatic and autonomous methodology to discover taxonomies of terms from the Web and represent retrieved web documents into a meaningful organization. Moreover, a new method for lexicalizations and synonyms discovery is also introduced. The obtained results can be very useful for easing the access to web resources of any medical domain or creating ontological representations of knowledge.

This paper proposes a selection of knowledge technologies for health care planning and decision support in regional-level management of Slovenian public health care. Data mining and statistical techniques were used to analyze databases collected by a regional Public Heath Institute. Specifically, we addressed the problem of directing patients from primary health care centers to specialists. Decision support tools were used for resource modeling in terms of availability and accessibility of public health services for the population. Specifically, we analyzed organisational aspects of public health resources in one Sovenian region (Celje) with the goal to identify the areas that are atypical in terms of availability and accessibility of public health services.

We present a novel Hierarchical Evolutionary Divide and Conquer method for automated, long-term planning of dietary menus. Dietary plans have to satisfy multiple numerical constraints (Reference Daily Intakes and balance on a daily and weekly basis) as well as criteria on the harmony (variety, contrast, color, appeal) of the components. Our multi-level approach solves problems via the decomposition of the search space and uses good solutions for sub-problems on higher levels of the hierarchy. Multi-Objective Genetic Algorithms are used on each level to create nutritionally adequate menus with a linear fitness combination extended with rule-based assessment. We also apply case-based initialization for starting the Genetic Algorithms from a better position of the search space. Results show that this combined strategy can cope with strict numerical constraints in a properly chosen algorithmic setup.

In the context of patients hospitalized in intensive care units, we would like to predict the evolution of the patient’s condition. We hypothesis that human-computer collaboration could help with the definition of signatures representative of specific situations. We have defined a multi-agent system () to support this assumption. Preliminary results are presented to support our assumption.

We have developed an algorithm for triaging acute pediatric abdominal pain in the Emergency Department using the discovery-driven approach. This algorithm is embedded into the MET-AP (Mobile Emergency Triage – Abdominal Pain) system – a clinical decision support system that assists physicians in making emergency triage decisions. In this paper we describe experimental evaluation of several data mining methods (inductive learning, case-based reasoning and Bayesian reasoning) and results leading to the selection of the rule-based algorithm.

In medicine, data mining methods such as Decision Tree Induction (DTI) can be trained for extracting rules to predict the outcomes of new patients. However, incompleteness and high dimensionality of stored data are a problem. Canonical Correlation Analysis (CCA) can be used prior to DTI as a dimension reduction technique to preserve the character of the original data by omitting non-essential data. In this study, data from 3949 breast cancer patients were analysed. Raw data were cleaned by running a set of logical rules. Missing values were replaced using the Expectation Maximization algorithm. After dimension reduction with CCA, DTI was employed to analyse the resulting dataset. The validity of the predictive model was confirmed by ten-fold cross validation and the effect of pre-processing was analysed by applying DTI to data without pre-processing. Replacing missing values and using CCA for data reduction dramatically reduced the size of the resulting tree and increased the accuracy of the prediction of breast cancer recurrence.

This paper describes the architecture and application of EpiXCS, a learning classifier system that uses reinforcement learning and the genetic algorithm to discover rule-based knowledge in epidemiologic surveillance databases. EpiXCS implements several additional features that tailor the XCS paradigm to the demands of epidemiologic data and users who are not familiar with learning classifier systems. These include a workbench-style interface for visualization and parameterization and the use of clinically meaningful evaluation metrics. EpiXCS has been applied to a large surveillance database, and shown to discover classification rules similarly to See5, a well-known decision tree inducer.