Role of Nursing Informatics in the Automation of Pneumonia Quality Measure Data Elements.

Core measures are standard metrics to reflect the processes of care provided by hospitals. Hospitals in the United States are expected to extract data from electronic health records, automated computation of core measures, and electronic submission of the quality measures data. Traditional manual calculation processes are time intensive and susceptible to error. Automated calculation has the potential to provide timely, accurate information, which could guide quality-of-care decisions, but this vision has yet to be achieved. In this study, nursing informaticists and data analysts implemented a method to automatically extract data elements from electronic health records to calculate a core measure. We analyzed the sensitivity, specificity, and accuracy of core measure data elements extracted via SQL query and compared the results to manually extracted data elements. This method achieved excellent performance for the structured data elements but was less efficient for semistructured and unstructured elements. We analyzed challenges in automating the calculation of quality measures and proposed a rule-based (hybrid) approach for semistructured and unstructured data elements.

Exploiting missing clinical data in Bayesian network modeling for predicting medical problems.

When machine learning algorithms are applied to data collected during the course of clinical care, it is generally accepted that the data has not been consistently collected. The absence of expected data elements is common and the mechanism through which a data element is missing often involves the clinical relevance of that data element in a specific patient. Therefore, the absence of data may have information value of its own. In the process of designing an application intended to support a medical problem list, we have studied whether the "missingness" of clinical data can provide useful information in building prediction models. In this study, we experimented with four methods of treating missing values in a clinical data set-two of them explicitly model the absence or "missingness" of data. Each of these data sets were used to build four different kinds of Bayesian classifiers-a naive Bayes structure, a human-composed network structure, and two networks based on structural learning algorithms. We compared the performance between groups with and without explicit models of missingness using the area under the ROC curve. The results showed that in most cases the classifiers trained using the explicit missing value treatments performed better. The result suggests that information may exist in "missingness" itself. Thus, when designing a decision support system, we suggest one consider explicitly representing the presence/absence of data in the underlying logic.

PTXT Finder - an application for finding appropriate EHR data elements for data analysis using cross referencing information sources.

PTXT Finder was developed to reduce the manual efforts necessary to map from clinical variables in a decision support system to data elements in an EHR. The descriptions in a data dictionary may be inadequate for pinpointing data elements that represent a clinical variable. Semantics implied in taxonomy and real usage of the element are two important supporting information sources. PTXT Finder provides description, hierarchy, and statistics displays so users can cross-reference among these views.

Data preparation framework for preprocessing clinical data in data mining.

Electronic health records are designed to provide online transactional data recording and reporting services that support the health care process. The characteristics of clinical data as it originates during the process of clinical documentation, including issues of data availability and complex representation models can make data mining applications challenging. Data preprocessing and transformation are required before one can apply data mining to clinical data. In this article, an approach to data preparation that utilizes information from the data, metadata and sources of medical knowledge is described. Heuristic rules and policies are defined for these three types of supporting information. Compared with an entirely manual process for data preparation, this approach can potentially reduce manual work by achieving a degree of automation in the rule creation and execution. A pilot experiment demonstrates that data sets created through this approach lead to a better model learning results than a fully manual process.