Identifying the ingredients of hydrous arc magmas: insights from Mt Lamington, Papua New Guinea.

Volcanism is the surface expression of magma intrusion, crystallization, assimilation and hybridization processes operating throughout the crust over a range of time periods. Many magmas, including those erupted at subduction zones, have complex textures that reflect these processes. Here, we use textural and geochemical characteristics of calcic amphiboles to help identify multiple ingredients of subduction zone magmatism at Mt Lamington volcano, Papua New Guinea. Our approach uses existing trace element partitioning schemes to calculate the compositions of amphibole equilibrium melts (AEMs). The AEM compositions show that Mt Lamington andesites and plutonic enclaves are dominated by fractionation of amphibole + plagioclase + biotite, with assimilation of plagioclase and zircon. Magnesiohastingsite crystals in the andesite and diktytaxitic mafic enclaves reflect multiple episodes of recharge by more primitive, geochemically variable melts. The andesite also contains clots with rounded grains and melt on grain boundaries. These features indicate slow crystallization, and the retention of melt films could significantly enhance the potential for remobilization of crystals by infiltrating melts or during magma mixing. Variations in crystallization conditions could thus significantly affect the mush microstructure. We suggest that this could result in a significant bias of the volcanic record towards the preferential incorporation of more slowly cooled plutonic material from the lower crust or from more thermally mature plumbing systems. This article is part of the Theo Murphy meeting issue 'Magma reservoir architecture and dynamics'.

A new prior for bayesian anomaly detection: application to biosurveillance.

OBJECTIVES: Bayesian anomaly detection computes posterior probabilities of anomalous events by combining prior beliefs and evidence from data. However, the specification of prior probabilities can be challenging. This paper describes a Bayesian prior in the context of disease outbreak detection. The goal is to provide a meaningful, easy-to-use prior that yields a posterior probability of an outbreak that performs at least as well as a standard frequentist approach. If this goal is achieved, the resulting posterior could be usefully incorporated into a decision analysis about how to act in light of a possible disease outbreak. METHODS: This paper describes a Bayesian method for anomaly detection that combines learning from data with a semi-informative prior probability over patterns of anomalous events. A univariate version of the algorithm is presented here for ease of illustration of the essential ideas. The paper describes the algorithm in the context of disease-outbreak detection, but it is general and can be used in other anomaly detection applications. For this application, the semi-informative prior specifies that an increased count over baseline is expected for the variable being monitored, such as the number of respiratory chief complaints per day at a given emergency department. The semi-informative prior is derived based on the baseline prior, which is estimated from using historical data. RESULTS: The evaluation reported here used semi-synthetic data to evaluate the detection performance of the proposed Bayesian method and a control chart method, which is a standard frequentist algorithm that is closest to the Bayesian method in terms of the type of data it uses. The disease-outbreak detection performance of the Bayesian method was statistically significantly better than that of the control chart method when proper baseline periods were used to estimate the baseline behavior to avoid seasonal effects. When using longer baseline periods, the Bayesian method performed as well as the control chart method. The time complexity of the Bayesian algorithm is linear in the number of the observed events being monitored, due to a novel, closed-form derivation that is introduced in the paper. CONCLUSIONS: This paper introduces a novel prior probability for Bayesian outbreak detection that is expressive, easy-to-apply, computationally efficient, and performs as well or better than a standard frequentist method.

Discovery of causal relationships in a gene-regulation pathway from a mixture of experimental and observational DNA microarray data.

This paper reports the methods and results of a computer-based search for causal relationships in the gene-regulation pathway of galactose metabolism in the yeast Saccharomyces cerevisiae. The search uses recently published data from cDNA microarray experiments. A Bayesian method was applied to learn causal networks from a mixture of observational and experimental gene-expression data. The observational data were gene-expression levels obtained from unmanipulated "wild-type" cells. The experimental data were produced by deleting ("knocking out") genes and observing the expression levels of other genes. Causal relations predicted from the analysis on 36 galactose gene pairs are reported and compared with the known galactose pathway. Additional exploratory analyses are also reported.

A simple algorithm for identifying negated findings and diseases in discharge summaries.

Narrative reports in medical records contain a wealth of information that may augment structured data for managing patient information and predicting trends in diseases. Pertinent negatives are evident in text but are not usually indexed in structured databases. The objective of the study reported here was to test a simple algorithm for determining whether a finding or disease mentioned within narrative medical reports is present or absent. We developed a simple regular expression algorithm called NegEx that implements several phrases indicating negation, filters out sentences containing phrases that falsely appear to be negation phrases, and limits the scope of the negation phrases. We compared NegEx against a baseline algorithm that has a limited set of negation phrases and a simpler notion of scope. In a test of 1235 findings and diseases in 1000 sentences taken from discharge summaries indexed by physicians, NegEx had a specificity of 94.5% (versus 85.3% for the baseline), a positive predictive value of 84.5% (versus 68.4% for the baseline) while maintaining a reasonable sensitivity of 77.8% (versus 88.3% for the baseline). We conclude that with little implementation effort a simple regular expression algorithm for determining whether a finding or disease is absent can identify a large portion of the pertinent negatives from discharge summaries.

Evaluation of negation phrases in narrative clinical reports.

OBJECTIVE: Automatically identifying findings or diseases described in clinical textual reports requires determining whether clinical observations are present or absent. We evaluate the use of negation phrases and the frequency of negation in free-text clinical reports. METHODS: A simple negation algorithm was applied to ten types of clinical reports (n=42,160) dictated during July 2000. We counted how often each of 66 negation phrases was used to mark a clinical observation as absent. Physicians read a random sample of 400 sentences, and precision was calculated for the negation phrases. We measured what proportion of clinical observations were marked as absent. RESULTS: The negation algorithm was triggered by sixty negation phrases with just seven of the phrases accounting for 90% of the negations. The negation phrases received an overall precision of 97%, with "not" earning the lowest precision of 63%. Between 39% and 83% of all clinical observations were identified as absent by the negation algorithm, depending on the type of report analyzed. The most frequently used clinical observations were negated the majority of the time. CONCLUSION: Because clinical observations in textual patient records are frequently negated, identifying accurate negation phrases is important to any system processing these reports.

Predicting ICU mortality: a comparison of stationary and nonstationary temporal models.

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.

Causal discovery from medical textual data.

Medical records usually incorporate investigative reports, historical notes, patient encounters or discharge summaries as textual data. This study focused on learning causal relationships from intensive care unit (ICU) discharge summaries of 1611 patients. Identification of the causal factors of clinical conditions and outcomes can help us formulate better management, prevention and control strategies for the improvement of health care. For causal discovery we applied the Local Causal Discovery (LCD) algorithm, which uses the framework of causal Bayesian Networks to represent causal relationships among model variables. LCD takes as input a dataset and outputs causes of the form variable Y causally influences variable Z. Using the words that occur in the discharge summaries as attributes for input, LCD output 8 purported causal relationships. The relationships ranked as most probable subjectively appear to be most causally plausible.

A study in causal discovery from population-based infant birth and death records.

In the domain of medicine, identification of the causal factors of diseases and outcomes, helps us formulate better management, prevention and control strategies for the improvement of health care. With the goal of exploring, evaluating and refining techniques to learn causal relationships from observational data, such as data routinely collected in healthcare settings, we focused on investigating factors that may contribute causally to infant mortality in the United States. We used the U.S. Linked Birth/Infant Death dataset for 1991 with more than four million records and about 200 variables for each record. Our sample consisted of 41,155 records randomly selected from the whole dataset. Each record had maternal, paternal and child factors and the outcome at the end of the first year--whether the infant survived or not. For causal discovery we used a modified Local Causal Discovery (LCD2) algorithm, which uses the framework of causal Bayesian Networks to represent causal relationships among model variables. LCD2 takes as input a dataset and outputs causes of the form variable X causes variable Y. Using the infant birth and death dataset as input, LCD2 output nine purported causal relationships. Eight out of the nine relationships seem plausible. Even though we have not yet discovered a clinically novel causal link, we plan to look for novel causal pathways using the full sample after refining the algorithm and developing a more efficient implementation.

Identifying patient subgroups with simple Bayes'.

Medical records can form the basis of retrospective studies, be used to evaluate hospital practices and guidelines, and provide examples for teaching medicine. Each of these tasks presumes the ability to accurately identify patient subgroups. We describe a method for selecting patient subgroups based on the text of their medical records and demonstrate its effectiveness. We also describe a modification of the basic system that does not assume the existence of a preclassified training set, and illustrate its effectiveness in one retrieval task.

An experiment comparing lexical and statistical methods for extracting MeSH terms from clinical free text.

OBJECTIVE: A primary goal of the University of Pittsburgh's 1990-94 UMLS-sponsored effort was to develop and evaluate PostDoc (a lexical indexing system) and Pindex (a statistical indexing system) comparatively, and then in combination as a hybrid system. Each system takes as input a portion of the free text from a narrative part of a patient's electronic medical record and returns a list of suggested MeSH terms to use in formulating a Medline search that includes concepts in the text. This paper describes the systems and reports an evaluation. The intent is for this evaluation to serve as a step toward the eventual realization of systems that assist healthcare personnel in using the electronic medical record to construct patient-specific searches of Medline. DESIGN: The authors tested the performances of PostDoc, Pindex, and a hybrid system, using text taken from randomly selected clinical records, which were stratified to include six radiology reports, six pathology reports, and six discharge summaries. They identified concepts in the clinical records that might conceivably be used in performing a patient-specific Medline search. Each system was given the free text of each record as an input. The extent to which a system-derived list of MeSH terms captured the relevant concepts in these documents was determined based on blinded assessments by the authors. RESULTS: PostDoc output a mean of approximately 19 MeSH terms per report, which included about 40% of the relevant report concepts. Pindex output a mean of approximately 57 terms per report and captured about 45% of the relevant report concepts. A hybrid system captured approximately 66% of the relevant concepts and output about 71 terms per report. CONCLUSION: The outputs of PostDoc and Pindex are complementary in capturing MeSH terms from clinical free text. The results suggest possible approaches to reduce the number of terms output while maintaining the percentage of terms captured, including the use of UMLS semantic types to constrain the output list to contain only clinically relevant MeSH terms.

The impact of modeling the dependencies among patient findings on classification accuracy and calibration.

We present a new Bayesian classifier for computer-aided diagnosis. The new classifier builds upon the naive-Bayes classifier, and models the dependencies among patient findings in an attempt to improve its performance, both in terms of classification accuracy and in terms of calibration of the estimated probabilities. This work finds motivation in the argument that highly calibrated probabilities are necessary for the clinician to be able to rely on the model's recommendations. Experimental results are presented, supporting the conclusion that modeling the dependencies among findings improves calibration.

Temporal representation design principles: an assessment in the domain of liver transplantation.

Time modeling is an important aspect of medical decision-support systems engineering. At the core of effective time modeling lies the challenge of proper knowledge representation design. In this paper, we focus on two important principles for effective time-modeling languages: (a) hybrid temporal representation, and (b) dynamic temporal abstraction. To explore the significance of these design principles, we extend a previously-defined formalism (single-granularity modifiable temporal belief networks--MTBN-SGs) to accommodate multiple temporal granularities and dynamic query and domain-specific model creation. We call the new formalism multiple-granularity MTBNs (MTBN-MGs). We develop a prototype system for modeling aspects of liver transplantation and analyze the resulting model with respect to its representation power, representational tractability, and inferential tractability. Our experiment demonstrates that the design of formalisms is crucial for effective time modeling. In particular: (i) Hybrid temporal representation is a desirable property of time-modeling languages because it makes knowledge acquisition easier, and increases representational tractability. (ii) Dynamic temporal abstraction improves inferential and representational tractability significantly. We discuss a high-level procedure for extending existing languages to incorporate hybrid temporal representation and dynamic temporal abstraction.

Using computer modeling to help identify patient subgroups in clinical data repositories.

OBJECTIVE: The ability to accurately and efficiently identify patient cases of interest in a hospital information system has many important clinical, research, educational and administrative uses. The identification of cases of interest sometimes can be difficult. This paper describes a two-stage method for searching for cases of interest. DESIGN: First, a Boolean search is performed using coded database variables. The user classifies the retrieved cases as being of interest or not. Second, based on the user-classified cases, a computer model of the patient cases of interest is constructed. The model is then used to help locate additional cases. These cases provide an augmented training set for constructing a new computer model of the cases of interest. This cycle of modeling and user classification continues until halted by the user. MEASUREMENTS: This paper describes a pilot study in which this method is used to identify the records of patients who have venous thrombosis. RESULTS: The results indicate that computer modeling enhances the identification of patient cases of interest.

Representing and developing temporally abstracted knowledge as a means towards facilitating time modeling in medical decision-support systems.

The utilization of the appropriate level of temporal abstraction is an important aspect of time modeling. We discuss some aspects of the relation of temporal abstraction to important knowledge engineering parameters such as model correctness, ease of model specification, knowledge availability, query completeness, inference tractability, and semantic clarity. We propose that versatile and efficient time-modeling formalisms should encompass ways to represent and reason at more than one level of abstraction, and we discuss such a hybrid formalism. Although many research efforts have concentrated on the automation of specific temporal abstractions, much research needs to be done in understanding and developing provably optimal abstractions. We provide an initial framework for studying this problem in a manner that is independent of the particular problem domain and knowledge representation, and suggest several research challenges that appear worth pursuing.

An evaluation of machine-learning methods for predicting pneumonia mortality.

This paper describes the application of eight statistical and machine-learning methods to derive computer models for predicting mortality of hospital patients with pneumonia from their findings at initial presentation. The eight models were each constructed based on 9847 patient cases and they were each evaluated on 4352 additional cases. The primary evaluation metric was the error in predicted survival as a function of the fraction of patients predicted to survive. This metric is useful in assessing a model's potential to assist a clinician in deciding whether to treat a given patient in the hospital or at home. We examined the error rates of the models when predicting that a given fraction of patients will survive. We examined survival fractions between 0.1 and 0.6. Over this range, each model's predictive error rate was within 1% of the error rate of every other model. When predicting that approximately 30% of the patients will survive, all the models have an error rate of less than 1.5%. The models are distinguished more by the number of variables and parameters that they contain than by their error rates; these differences suggest which models may be the most amenable to future implementation as paper-based guidelines.

A temporal analysis of QMR.

OBJECTIVE: To understand better the trade-offs of not incorporating explicit time in Quick Medical Reference (QMR), a diagnostic system in the domain of general internal medicine, along the dimensions of expressive power and diagnostic accuracy. DESIGN: The study was conducted in two phases. Phase I was a descriptive analysis of the temporal abstractions incorporated in QMR's terms. Phase II was a pseudo-prospective controlled experiment, measuring the effect of history and physical examination temporal content on the diagnostic accuracy of QMR. MEASUREMENTS: For each QMR finding that would fit our operational definition of temporal finding, several parameters describing the temporal nature of the finding were assessed, the most important ones being: temporal primitives, time units, temporal uncertainty, processes, and patterns. The history, physical examination, and initial laboratory results of 105 consecutive patients admitted to the Pittsburgh University Presbyterian Hospital were analyzed for temporal content and factors that could potentially influence diagnostic accuracy (these included: rareness of primary diagnosis, case length, uncertainty, spatial/causal information, and multiple diseases). RESULTS: 776 findings were identified as temporal. The authors developed an ontology describing the terms utilized by QMR developers to express temporal knowledge. The authors classified the temporal abstractions found in QMR in 116 temporal types, 11 temporal templates, and a temporal hierarchy. The odds of QMR's making a correct diagnosis in high temporal complexity cases is 0.7 the odds when the temporal complexity is lower, but this result is not statistically significant (95% confidence interval = 0.27-1.83). CONCLUSIONS: QMR contains extensive implicit time modeling. These results support the conclusion that the abstracted encoding of time in the medical knowledge of QMR does not induce a diagnostic performance penalty.

Human causal discovery from observational data.

Utilizing Bayesian belief networks as a model of causality, we examined medical students' ability to discover causal relationships from observational data. Nine sets of patient cases were generated from relatively simple causal belief networks by stochastic simulation. Twenty participants examined the data sets and attempted to discover the underlying causal relationships. Performance was poor in general, except at discovering the absence of a causal relationship. This work supports the potential for combining human and computer methods for causal discovery.

Representing CARE rules in a decision-theoretic formalism.

Improvement in the performance of reminder systems may be facilitated by the use of new representations. A decision-theoretic representation, for example, may enable a reminder system to represent and reason about the probabilities that a reminder will be a true or a false alarm and the relative utilities of these events. We extended a previously described decision-theoretic model to include such events. The model now represents explicitly the uncertainty, costs, and benefits of sending a reminder. We also extended the model to remove an assumption of reminder independence. As a step towards testing a hypothesis that this approach will support better performance than a rule-based approach, we analyzed a set of CARE rules and showed that our representation can represent these rules.

Progress in the development of trauma systems in the United States. Results of a national survey.

OBJECTIVE: To examine the status of trauma system development and key structural and operational characteristics of these systems. DESIGN AND SETTING: National survey of trauma systems with enabling state statute, regulation, or executive orders and for which designated trauma centers were present. PARTICIPANTS: Trauma system administrators and directors of 37 state and regional organizations that had legal authority to administer trauma systems, which represented a response rate of 90.2%. MAIN OUTCOME MEASURES: Trauma system components that had been implemented or were under development. RESULTS: From 1988 to 1993, the number of states meeting one set of criteria for a complete trauma system criteria increased from two to five. The most common deficiency in establishing trauma systems was failure to limit the number of designated trauma centers based on community need. Although most existing trauma systems have developed formal processes for designating trauma centers, prehospital triage protocols to allow hospital bypass, and centralized trauma registries, several systems lack standardized policies for interhospital transfer and systemwide evaluation. CONCLUSION: State and regional organizations have accomplished a great deal but still have substantial work ahead in developing comprehensive trauma systems. Research is needed to better understand the relationship between trauma volume and outcomes of care as well as the impact of trauma system structure and operational characteristics on care delivery. Improved measures of patient outcome are also needed so that effective system evaluation can take place.