Selective sampling to overcome skewed a priori probabilities with neural networks. (49/4007)

Highly skewed a priori probabilities present challenges for researchers developing medical decision aids due to a lack of information on the rare outcome of interest. This paper attempts to overcome this obstacle by artificially increasing the mortality rate of the training sets. A weight pruning technique called weight-elimination is also applied to this coronary artery bypass grafting (CABG) database to assess its impact on the artificial neural network's (ANN) performance. The results showed that increasing the mortality rate improved the sensitivity rates at the cost of the other performance measures, and the weight-elimination cost function improved the sensitivity rate without seriously affecting the other performance measures.  (+info)

A quality and safety framework for point-of-care clinical guidelines. (50/4007)

The electronic dissemination of medical knowledge in the form of executable clinical guidelines and decision support systems must be accompanied by comprehensive methods for ensuring the quality of their knowledge content and their safety in use. This paper outlines a set of quality and safety requirements, and reviews three current guideline technologies, the Arden Syntax, GLIF and PROforma, against these requirements. The approaches used in these technologies have different strengths, and we propose a general framework for ensuring quality and safety that combines them. This framework brings together the normal documentation standards of medical publishing, rigorous design methods from software engineering, and active safety management techniques from artificial intelligence.  (+info)

Implementing clinical practice guidelines while taking account of changing evidence: ATHENA DSS, an easily modifiable decision-support system for managing hypertension in primary care. (51/4007)

This paper describes the ATHENA Decision Support System (DSS), which operationalizes guidelines for hypertension using the EON architecture. ATHENA DSS encourages blood pressure control and recommends guideline-concordant choice of drug therapy in relation to comorbid diseases. ATHENA DSS has an easily modifiable knowledge base that specifies eligibility criteria, risk stratification, blood pressure targets, relevant comorbid diseases, guideline-recommended drug classes for patients with comorbid disease, preferred drugs within each drug class, and clinical messages. Because evidence for best management of hypertension evolves continually, ATHENA DSS is designed to allow clinical experts to customize the knowledge base to incorporate new evidence or to reflect local interpretations of guideline ambiguities. Together with its database mediator Athenaeum, ATHENA DSS has physical and logical data independence from the legacy Computerized Patient Record System (CPRS) supplying the patient data, so it can be integrated into a variety of electronic medical record systems.  (+info)

Building knowledge in a complex preterm birth problem domain. (52/4007)

Data mining methods used a racially diverse sample (n = 19,970) of pregnant women and 1,622 variables that were collected in Duke's TMR electronic patient record over a 10-year period. Different statistical and data mining methods were similar when compared using receiver operating characteristic (ROC) curves. Best results found that seven demographic variables yielded .72 and addition of hundreds of other clinical variables added only .03 to the area under the curve (AUC). Similar results across methods suggest that results were data-driven and not method-dependent, and that demographic variables may offer a small set of parsimonious variables with predictive accuracy in a racially diverse population. Work to determine relevant variables for improved predictive accuracy is ongoing.  (+info)

Discovery of predictive models in an injury surveillance database: an application of data mining in clinical research. (53/4007)

A new, evolutionary computation-based approach to discovering prediction models in surveillance data was developed and evaluated. This approach was operationalized in EpiCS, a type of learning classifier system specially adapted to model clinical data. In applying EpiCS to a large, prospective injury surveillance database, EpiCS was found to create accurate predictive models quickly that were highly robust, being able to classify > 99% of cases early during training. After training, EpiCS classified novel data more accurately (p < 0.001) than either logistic regression or decision tree induction (C4.5), two traditional methods for discovering or building predictive models.  (+info)

Predicting ICU mortality: a comparison of stationary and nonstationary temporal models. (54/4007)

OBJECTIVE: This study evaluates the effectiveness of the stationarity assumption in predicting the mortality of intensive care unit (ICU) patients at the ICU discharge. DESIGN: This is a comparative study. A stationary temporal Bayesian network learned from data was compared to a set of (33) nonstationary temporal Bayesian networks learned from data. A process observed as a sequence of events is stationary if its stochastic properties stay the same when the sequence is shifted in a positive or negative direction by a constant time parameter. The temporal Bayesian networks forecast mortalities of patients, where each patient has one record per day. The predictive performance of the stationary model is compared with nonstationary models using the area under the receiver operating characteristics (ROC) curves. RESULTS: The stationary model usually performed best. However, one nonstationary model using large data sets performed significantly better than the stationary model. CONCLUSION: Results suggest that using a combination of stationary and nonstationary models may predict better than using either alone.  (+info)

Knowledge representation forms for data mining methodologies as applied in thoracic surgery. (55/4007)

Typical ways of disseminating and using results of clinical research are scientific journals and reports. Presentation forms are condensed and comprehensible mainly to the experts following the specific topics. A vast amount of information remains unutilized due to the complex form of presenting the knowledge. Subject of this research is to explore possibilities of representation and also visualization of the results obtained using data mining methodologies. The intention is to formulate more than scientific ways to communicate facts that are of interest for the clinicians, medical students and even patients. Internet technologies as already widely established media support knowledge representation forms such as hypertext documents and structured knowledge components. The "Assist Me" decision support system for surgical treatment of cardiac patients integrates several forms of data mining and representation methodologies. We are showing a feasibility study in which scientific outcomes were forwarded to a broad group of potential users.  (+info)

Knowledge acquisition to qualify Unified Medical Language System interconceptual relationships. (56/4007)

Adding automatically relations between concepts from a database to a knowledge base such as the Unified Medical Language System can be very useful to increase the consistency of the latter one. But the transfer of qualified relationships is more interesting. The most important interest of these new acquisitions is that the UMLS became more compliant and medically pertinent to be used in different medical applications. This paper describes the possibility to inherit automatically medical inter-conceptual relationships qualifiers from a disease description included into a database and to integrate them into the UMLS knowledge base. The paper focuses on the transmission of knowledge from a French medical database to an English one.  (+info)