Search

Predictive models can be useful in predicting patient outcomes under uncertainty. Many algorithms employ "population" methods, which optimize a single model to perform well on average over an entire population, but the model may perform poorly on some patients. Personalized methods optimize predictive performance for each patient by tailoring the model to the individual. We present a new personalized method based on decision trees: the Personalized Decision Path using a Bayesian score (PDP-Bay). Performance on eight synthetic, genomic, and clinical datasets was compared to that of decision trees and a previously described personalized decision path method in terms of area under the ROC curve (AUC) and expected calibration error (ECE). Model complexity was measured by average path length. The PDP-Bay model outperformed the decision tree in terms of both AUC and ECE. The results support the conclusion that personalization may achieve better predictive performance and produce simpler models than population approaches., (©2020 AMIA - All rights reserved.)

Poor electronic medical record (EMR) usability is detrimental to both clinicians and patients. A better EMR would provide concise, context sensitive patient data, but doing so entails the difficult task of knowing which data are relevant. To determine the relevance of patient data in different contexts, we collect and model the information seeking behavior of clinicians using a learning EMR (LEMR) system. Sufficient data were collected to train predictive models for 80 different targets (e.g., glucose level, heparin administration) and 27 of them had AUROC values of greater than 0.7. These results are encouraging considering the high variation in information seeking behavior (intraclass correlation 0.40). We plan to apply these models to a new set of patient cases and adapt the LEMR interface to highlight relevant patient data, and thus provide concise, context sensitive data.

The Accrual to Clinical Trials (ACT) network is a federated network of sites from the National Clinical and Translational Science Award (CTSA) Consortium that has been created to significantly increase participant accrual to multi-site clinical trials. The ACT network represents an unprecedented collaboration among diverse CTSA sites. The network has created governance and regulatory frameworks and a common data model to harmonize electronic health record (EHR) data, and deployed a set of Informatics for Integrating Biology and the Bedside (i2b2) data repositories that are linked by the Shared Health Research Information Network (SHRINE) platform. It provides investigators the ability to query the network in real time and to obtain aggregate counts of patients who meet clinical trial inclusion and exclusion criteria from sites across the United States. The ACT network infrastructure provides a basis for cohort discovery and for developing new informatics tools to identify and recruit participants for multi-site clinical trials.

Bayesian network (BN) structure learning algorithms are almost always designed to recover the structure that models the relationships that are shared by the instances in a population . While accurately learning such population-wide Bayesian networks is useful, learning Bayesian networks that are specific to each instance is often important as well. For example, to understand and treat a patient (instance), it is critical to understand the specific causal mechanisms that are operating in that particular patient. We introduce an instance-specific BN structure learning method that searches the space of Bayesian networks to build a model that is specific to an instance by guiding the search based on attributes of the given instance (e.g., patient symptoms, signs, lab results, and genotype). The structure discovery performance of the proposed method is compared to an existing state-of-the-art BN structure learning method, namely an implementation of the Greedy Equivalence Search algorithm called FGES, using both simulated and real data. The results show that the proposed method improves the precision of the model structure that is output, when compared to GES, especially for those variables that exhibit context-specific independence.

Many real datasets contain values missing not at random (MNAR). In this scenario, investigators often perform list-wise deletion, or delete samples with any missing values, before applying causal discovery algorithms. List-wise deletion is a sound and general strategy when paired with algorithms such as FCI and RFCI, but the deletion procedure also eliminates otherwise good samples that contain only a few missing values. In this report, we show that we can more efficiently utilize the observed values with test-wise deletion while still maintaining algorithmic soundness. Here, test-wise deletion refers to the process of list-wise deleting samples only among the variables required for each conditional independence (CI) test used in constraint-based searches. Test-wise deletion therefore often saves more samples than list-wise deletion for each CI test, especially when we have a sparse underlying graph. Our theoretical results show that test-wise deletion is sound under the justifiable assumption that none of the missingness mechanisms causally affect each other in the underlying causal graph. We also find that FCI and RFCI with test-wise deletion outperform their list-wise deletion and imputation counterparts on average when MNAR holds in both synthetic and real data.

Eye-tracking is a valuable research tool that is used in laboratory and limited field environments. We take steps toward developing methods that enable widespread adoption of eye-tracking and its real-time application in clinical decision support. Eye-tracking will enhance awareness and enable intelligent views, more precise alerts, and other forms of decision support in the Electronic Medical Record (EMR). We evaluated a low-cost eye-tracking device and found the device's accuracy to be non-inferior to a more expensive device. We also developed and evaluated an automatic method for mapping eye-tracking data to interface elements in the EMR (e.g., a displayed laboratory test value). Mapping was 88% accurate across the six participants in our experiment. Finally, we piloted the use of the low-cost device and the automatic mapping method to label training data for a Learning EMR (LEMR) which is a system that highlights the EMR elements a physician is predicted to use.

A primary goal of precision medicine is to identify patient subgroups based on their characteristics (e.g., comorbidities or genes) with the goal of designing more targeted interventions. While network visualization methods such as Fruchterman-Reingold have been used to successfully identify such patient subgroups in small to medium sized data sets, they often fail to reveal comprehensible visual patterns in large and dense networks despite having significant clustering. We therefore developed an algorithm called ExplodeLayout , which exploits the existence of significant clusters in bipartite networks to automatically "explode" a traditional network layout with the goal of separating overlapping clusters, while at the same time preserving key network topological properties that are critical for the comprehension of patient subgroups. We demonstrate the utility of ExplodeLayout by visualizing a large dataset extracted from Medicare consisting of readmitted hip-fracture patients and their comorbidities, demonstrate its statistically significant improvement over a traditional layout algorithm, and discuss how the resulting network visualization enabled clinicians to infer mechanisms precipitating hospital readmission in specific patient subgroups.

The Breast Imaging Reporting and Data System (BI-RADS) was developed to reduce variation in the descriptions of findings. Manual analysis of breast radiology report data is challenging but is necessary for clinical and healthcare quality assurance activities. The objective of this study is to develop a natural language processing (NLP) system for automated BI-RADS categories extraction from breast radiology reports. We evaluated an existing rule-based NLP algorithm, and then we developed and evaluated our own method using a supervised machine learning approach. We divided the BI-RADS category extraction task into two specific tasks: (1) annotation of all BI-RADS category values within a report, (2) classification of the laterality of each BI-RADS category value. We used one algorithm for task 1 and evaluated three algorithms for task 2. Across all evaluations and model training, we used a total of 2159 radiology reports from 18 hospitals, from 2003 to 2015. Performance with the existing rule-based algorithm was not satisfactory. Conditional random fields showed a high performance for task 1 with an F-1 measure of 0.95. Rules from partial decision trees (PART) algorithm showed the best performance across classes for task 2 with a weighted F-1 measure of 0.91 for BIRADS 0-6, and 0.93 for BIRADS 3-5. Classification performance by class showed that performance improved for all classes from Naïve Bayes to Support Vector Machine (SVM), and also from SVM to PART. Our system is able to annotate and classify all BI-RADS mentions present in a single radiology report and can serve as the foundation for future studies that will leverage automated BI-RADS annotation, to provide feedback to radiologists as part of a learning health system loop., (Copyright © 2017. Published by Elsevier Inc.)

Background: Patient matching is a key barrier to achieving interoperability. Patient demographic elements must be consistently collected over time and region to be valuable elements for patient matching., Objectives: We sought to determine what patient demographic attributes are collected at multiple institutions in the United States and see how their availability changes over time and across clinical sites., Methods: We compiled a list of 36 demographic elements that stakeholders previously identified as essential patient demographic attributes that should be collected for the purpose of linking patient records. We studied a convenience sample of 9 health care systems from geographically distinct sites around the country. We identified changes in the availability of individual patient demographic attributes over time and across clinical sites., Results: Several attributes were consistently available over the study period (2005-2014) including last name (99.96%), first name (99.95%), date of birth (98.82%), gender/sex (99.73%), postal code (94.71%), and full street address (94.65%). Other attributes changed significantly from 2005-2014: Social security number (SSN) availability declined from 83.3% to 50.44% (p<0.0001). Email address availability increased from 8.94% up to 54% availability (p<0.0001). Work phone number increased from 20.61% to 52.33% (p<0.0001)., Conclusions: Overall, first name, last name, date of birth, gender/sex and address were widely collected across institutional sites and over time. Availability of emerging attributes such as email and phone numbers are increasing while SSN use is declining. Understanding the relative availability of patient attributes can inform strategies for optimal matching in healthcare.

Medical errors remain a significant problem in healthcare. This paper investigates a data-driven outlier-based monitoring and alerting framework that uses data in the Electronic Medical Records (EMRs) repositories of past patient cases to identify any unusual clinical actions in the EMR of a current patient. Our conjecture is that these unusual clinical actions correspond to medical errors often enough to justify their detection and alerting. Our approach works by using EMR repositories to learn statistical models that relate patient states to patient-management actions. We evaluated this approach on the EMR data for 24,658 intensive care unit (ICU) patient cases. A total of 16,500 cases were used to train statistical models for ordering medications and laboratory tests given the patient state summarizing the patient's clinical history. The models were applied to a separate test set of 8158 ICU patient cases and used to generate alerts. A subset of 240 alerts generated by the models were evaluated and assessed by eighteen ICU clinicians. The overall true positive rates for the alerts (TPARs) ranged from 0.44 to 0.71. The TPAR for medication order alerts specifically ranged from 0.31 to 0.61 and for laboratory order alerts from 0.44 to 0.75. These results support outlier-based alerting as a promising new approach to data-driven clinical alerting that is generated automatically based on past EMR data., (Copyright © 2016 Elsevier Inc. All rights reserved.)

Influenza is a yearly recurrent disease that has the potential to become a pandemic. An effective biosurveillance system is required for early detection of the disease. In our previous studies, we have shown that electronic Emergency Department (ED) free-text reports can be of value to improve influenza detection in real time. This paper studies seven machine learning (ML) classifiers for influenza detection, compares their diagnostic capabilities against an expert-built influenza Bayesian classifier, and evaluates different ways of handling missing clinical information from the free-text reports. We identified 31,268 ED reports from 4 hospitals between 2008 and 2011 to form two different datasets: training (468 cases, 29,004 controls), and test (176 cases and 1620 controls). We employed Topaz, a natural language processing (NLP) tool, to extract influenza-related findings and to encode them into one of three values: Acute, Non-acute, and Missing. Results show that all ML classifiers had areas under ROCs (AUC) ranging from 0.88 to 0.93, and performed significantly better than the expert-built Bayesian model. Missing clinical information marked as a value of missing (not missing at random) had a consistently improved performance among 3 (out of 4) ML classifiers when it was compared with the configuration of not assigning a value of missing (missing completely at random). The case/control ratios did not affect the classification performance given the large number of training cases. Our study demonstrates ED reports in conjunction with the use of ML and NLP with the handling of missing value information have a great potential for the detection of infectious diseases., (Copyright © 2015 Elsevier Inc. All rights reserved.)

Electronic medical records (EMRs) are capturing increasing amounts of data per patient. For clinicians to efficiently and accurately understand a patient's clinical state, better ways are needed to determine when and how to display EMR data. We built a prototype system that records how physicians view EMR data, which we used to train models that predict which EMR data will be relevant in a given patient. We call this approach a Learning EMR (LEMR). A physician used the prototype to review 59 intensive care unit (ICU) patient cases. We used the data-access patterns from these cases to train logistic regression models that, when evaluated, had AUROC values as high as 0.92 and that averaged 0.73, supporting that the approach is promising. A preliminary usability study identified advantages of the system and a few concerns about implementation. Overall, 3 of 4 ICU physicians were enthusiastic about features of the prototype.

Although a majority of 30-day readmissions of hip-fracture (HFx) patients in the elderly are caused by non-surgical complications, little is known about which specific combinations of comorbidities are associated with increased risk of readmission. We therefore used bipartite network analysis to explore the complex associations between 70 comorbidities (defined by hierarchal condition categories as critical in this population) and (a) cases consisting of all 2,316 HFx patients without hospital complications in the 2010 Medicare claims database who were re-admitted within 30 days of discharge, and (b) controls consisting of an equal number of matched HFx patients who were not readmitted for at least 90 days since discharge. A network-wide analysis revealed nine patient/comorbidity co-clusters, of which two had a significantly different proportion of cases compared to the rest of the data. A cluster-specific analysis of the most significant co-cluster revealed that a pair of comorbidities (Renal Failure and Diabetes with no Complications) within the co-cluster had significantly higher risk of 30-day readmission, whereas another pair of comorbidities (Renal Failure and Diabetes with Renal or Peripheral Circulatory Manifestations), despite having a relatively more serious comorbidity, did not confer a higher risk. This counter-intuitive result suggests that HFx patients with more serious comorbidities may have better follow-up that reduces the risk of 30-day readmission, whereas those with specific relatively less-serious comorbidities may have less stringent follow-up resulting in unanticipated incidents that precipitate readmission. These analyses reveal the strengths and limitations of bipartite networks for identifying hypotheses for complex phenomena related to readmissions, with the goal of improving follow-up care for patients with specific combinations of comorbidities.

Although influenza (flu) and respiratory syncytial virus (RSV) infections are extremely common in children under two years and resolve naturally, a subset develop severe disease resulting in hospitalization despite having no identifiable clinical risk factors. However, little is known about inherent host-specific genetic and immune mechanisms in this at-risk subpopulation. We therefore conducted a secondary analysis of statistically significant, differentially-expressed genes from a whole genome-wide case-control study of children less than two years of age hospitalized with flu or RSV, through the use of bipartite networks. The analysis revealed three clusters of cases common to both types of infection: core cases with high expression of genes in the network core implicated in hyperimmune responsiveness; periphery cases with lower expression of the same set of genes indicating medium-responsiveness; and control-like cases with a gene signature resembling that of the controls, indicating normal-responsiveness. These results provide testable hypotheses for at-risk subphenotypes and their respective pathophysiologies in both types of infections. We conclude by discussing alternate hypotheses for the results, and insights about how bipartite networks of gene expression across multiple phenotypes can help to identify complex patterns related to subphenotypes, with the translational goal of identifying targeted therapeutics.

Accurate disease classification and biomarker discovery remain challenging tasks in biomedicine. In this paper, we develop and test a practical approach to combining evidence from multiple models when making predictions using selective Bayesian model averaging of probabilistic rules. This method is implemented within a Bayesian Rule Learning system and compared to model selection when applied to twelve biomedical datasets using the area under the ROC curve measure of performance. Cross-validation results indicate that selective Bayesian model averaging statistically significantly outperforms model selection on average in these experiments, suggesting that combining predictions from multiple models may lead to more accurate quantification of classifier uncertainty. This approach would directly impact the generation of robust predictions on unseen test data, while also increasing knowledge for biomarker discovery and mechanisms that underlie disease.

Patient-specific models are constructed to take advantage of the particular features of the patient case of interest compared to commonly used population-wide models that are constructed to perform well on average on all cases. We introduce two patient-specific algorithms that are based on the decision tree paradigm. These algorithms construct a decision path specific for each patient of interest compared to a single population-wide decision tree with many paths that is applicable to all patients of interest that are constructed by standard algorithms. We applied the patient-specific algorithms to predict five different outcomes in clinical datasets. Compared to the population-wide CART decision tree the patient-specific decision path models had superior performance on area under the ROC curve (AUC) and had comparable performance on balanced accuracy. Our results provide support for patient-specific algorithms being a promising approach for predicting clinical outcomes.

Several intersecting host, vector, and environmental factors have led to a re-emergence of rickettsial diseases such as Mediterranean Spotted Fever (MSF), and Dermacentor spp.-borne necrosis-erythema lymphadenopathy (DEBONEL). Some rickettsiae produce diffuse endothelial infection and systemic microvascular leakage leading in some cases to high morbidity and mortality. Unfortunately, little is known about the molecular pathways triggered by these diseases in humans. We therefore analyzed how candidate cytokines co-occur across acutely-ill patients with either a localized (DEBONEL), or a systemic (MSF) form of rickettsiosis, using bipartite visual analytics. The results revealed a network core consisting of a small set of MSF patients exhibiting high expressions of cytokines implicated in microvascular leakage, endothelial repair, and pro-inflammatory immune responses, and a network periphery consisting of a mixture of MSF and DEBONEL patients with relatively lower overall cytokine expressions. These results provide evidence of pathways triggered by rickettsiae in humans, and a testable hypothesis for the mechanisms in a rickettsia-induced cytokine storm with the translational goal of identifying therapeutic targets.

We develop and evaluate a data-driven approach for detecting unusual (anomalous) patient-management actions using past patient cases stored in an electronic health record (EHR) system. Our hypothesis is that patient-management actions that are unusual with respect to past patients may be due to a potential error and that it is worthwhile to raise an alert if such a condition is encountered. We evaluate this hypothesis using data obtained from the electronic health records of 4,486 post-cardiac surgical patients. We base the evaluation on the opinions of a panel of experts. The results support that anomaly-based alerting can have reasonably low false alert rates and that stronger anomalies are correlated with higher alert rates.

This study explored the possibility that semantic distance metrics can be used to develop methods for auditing biomedical ontologies. We developed and tested an approach using the Foundational Model of Anatomy (FMA) and the body-structure taxonomy of SNOMED CT. We evaluated 190 class pairs in human anatomical structures using three semantic distance metrics: simple edge count, normalized path length, and information content. We applied principal component analysis (PCA) to study relationships between the semantic distance measurements so produced in FMA and SNOMED CT. We found that our application of PCA could detect significant discrepancies, but not necessarily outright mistakes, in the two ontologies. A review of discrepancies revealed that they often relate to multiple design perspectives employed in ontological definitions.

Genetic epidemiologists strive to determine the genetic profile of diseases. Epistasis is the interaction between two or more genes to affect phenotype. Due to the often non-linearity of the interaction, it is difficult to detect statistical patterns of epistasis. Combinatorial methods for detecting epistasis investigate a subset of combinations of genes without employing a search strategy. Therefore, they do not scale to handling the high-dimensional data found in genome-wide association studies (GWAS). We represent genome-phenome interactions using a Bayesian network rule, which is a specialized Bayesian network. We develop an efficient search algorithm to learn from data a high scoring rule that may contain two or more interacting genes. Our experimental results using synthetic data indicate that this algorithm detects interacting genes as well as a Bayesian network combinatorial method, and it is much faster. Our results also indicate that the algorithm can successfully learn genome-phenome relationships using a real GWAS dataset.

Developing methods to detect deviations from usual medical care may be useful in the development of automated clinical alerting systems to alert clinicians to treatment choices that warrant additional consideration. We developed a method for identifying deviations in medication administration in the intensive care unit that is based on learning logistic regression models from past patient data that when applied to current patient data identifies statistically unusual treatment decisions. The models predicted a total of 53 deviations for 6 medications on a set of 3000 patient cases. A set of 12 predicted deviations and 12 non-deviations was evaluated by a group of intensive care physicians. Overall, the predicted deviations were assessed to often warrant an alert and to be clinically useful, and furthermore, the frequency with which such alerts would be raised is not likely to be disruptive in a clinical setting.

This paper compares the predictive performance and efficiency of several machine-learning methods when applied to a genome-wide dataset on Alzheimer's disease that contains 312,318 SNP measurements on 1411 cases. In particular, a Bayesian algorithm is introduced and compared to several standard machine-learning methods. The results show that the Bayesian algorithm predicts outcomes comparably to the standard methods, and it requires less total training time. These results support the further development and evaluation of the Bayesian algorithm.

An important challenge in the analysis of single nucleotide polymorphism (SNP) data is the identification of SNPs that interact in a nonlinear fashion in their association with disease. Such epistatic interactions among genetic variants at multiple loci likely underlie the inheritance of common diseases. We have developed a novel method called the Bayesian combinatorial method (BCM) for detecting combination of genetic variants that are predictive of disease. When compared with the multifactor dimensionality reduction (MDR), a widely used combinatorial method, BCM has significantly greater power to detect interactions and is computationally more efficient.

An important step in the analysis of high-dimensional biomedical data is feature selection. Typically, a feature subset selected by a feature selection method is evaluated for relevance towards a task such as prediction or classification. Another important property of a feature selection method is stability that refers to robustness of the selected features to perturbations in the data. In biomarker discovery, for example, domain experts prefer a parsimonious subset of features that are relatively robust to slight changes in the data. We present a stability measure called the adjusted stability measure that computes robustness of a feature selection method with respect to random feature selection. This measure is useful for comparing the robustness of feature selection methods and is superior to similar measures that do not account for random feature selection. We demonstrate the application of this measure on a biomedical dataset.

The Empirical Proteomic Ontology Knowledge Base (EPO-KB) is an online database that represents current knowledge of biomarkers and contains associations between mass-to-charge (m/z) ratios of mass-spectrometry peaks to proteins. Such a database is a useful tool for identifying putative proteins associated with a m/z ratio. At present, EPO-KB contains data that have been extracted from 120 published research papers. It has been used in successful identification of a protein associated with a biomarker.

In order to increase compliance with The Joint Commission's Congestive Heart Failure Core Measures, a rule based clinical decision support system (CDSS) was developed and deployed at a community hospital in our health system. We evaluated the performance of the CDSS in identifying patients with primary congestive heart failure (CHF)and identified problems encountered with its introduction. Performance of the CDSS was compared against a manual review of records of patients with diagnosis of primary CHF. The CDSS had a sensitivity of 0.79 and PPV of 0.11. The CDSS issued multiple alerts for majority of the patients(74%). The number of alerts issued for patients without primary CHF was large, and for a majority of patients (63%) physicians did not respond to alerts the first time. The CDSS performed poorly and was eventually withdrawn but provided insight into a subsequently successful method for managing CHF.

Adverse drug reactions (ADRs) are a common cause of morbidity and mortality in the nursing home (NH) setting. Traditional non-automated mechanisms for ADR detection are time-consuming, costly, and fail to detect the majority of ADRs. We describe the implementation and pharmacist evaluation of a clinical event monitor using signals previously developed by our research team to detect potential ADRs in the NH. The overall positive predictive value (PPV) for all signals combined was 81% (54/67), with individual signal PPVs ranging from 0-100%. The PPVs were 53% (10/19) for the antidote signals category and 96% (44/46) for the laboratory/ medication combination signals category. The majority 75% (12/16) of the preventable ADRs were laboratory/medication combination signals. The results suggest that ADRs can be detected in the NH setting with a high degree of accuracy using a clinical event monitor that employs a set of signals derived by expert consensus.

Discretization acts as a variable selection method in addition to transforming the continuous values of the variable to discrete ones. Machine learning algorithms such as Support Vector Machines and Random Forests have been used for classification in high-dimensional genomic and proteomic data due to their robustness to the dimensionality of the data. We show that discretization can help improve significantly the classification performance of these algorithms as well as algorithms like Naïve Bayes that are sensitive to the dimensionality of the data.

Anomaly detection methods can be very useful in identifying unusual or interesting patterns in data. A recently proposed conditional anomaly detection framework extends anomaly detection to the problem of identifying anomalous patterns on a subset of attributes in the data. The anomaly always depends (is conditioned) on the value of remaining attributes. The work presented in this paper focuses on instance-based methods for detecting conditional anomalies. The methods rely on the distance metric to identify examples in the dataset that are most critical for detecting the anomaly. We investigate various metrics and metric learning methods to optimize the performance of the instance-based anomaly detection methods. We show the benefits of the instance-based methods on two real-world detection problems: detection of unusual admission decisions for patients with the community-acquired pneumonia and detection of unusual orders of an HPF4 test that is used to confirm Heparin induced thrombocytopenia - a life-threatening condition caused by the Heparin therapy.

Detecting deviations from usual medical care is valuable in identifying potentially concerning patient management events, both in real time and retrospectively. We describe a statistical method for identification of deviations in medication administration. The preliminary results reported here characterize the statistical properties of the identified deviations. Future research will investigate which deviations are clinically useful.

Anomaly detection methods can be very useful in identifying interesting or concerning events. In this work, we develop and examine new probabilistic anomaly detection methods that let us evaluate management decisions for a specific patient and identify those decisions that are highly unusual with respect to patients with the same or similar condition. The statistics used in this detection are derived from probabilistic models such as Bayesian networks that are learned from a database of past patient cases. We evaluate our methods on the problem of detection of unusual hospitalization patterns for patients with community acquired pneumonia. The results show very encouraging detection performance with 0.5 precision at 0.53 recall and give us hope that these techniques may provide the basis of intelligent monitoring systems that alert clinicians to the occurrence of unusual events or decisions.

In the United States, penicillin-resistant variants of the Tennessee (Tenn) (23 F)-4 clone account for a substantial proportion of the very-high-level penicillin-resistant (MIC 8 microg/ml) infections in the 7-valent pneumococcal protein conjugate vaccine (PCV 7) era. Serotype 19 A strains account for an increasing proportion of penicillin-nonsusceptible Streptococcus pneumoniae infections. Sequential transformations of the Tenn (23 F)-4 clone (penicillin MIC 0.1 microg/ml) were performed with four penicillin-nonsusceptible serotype 19 A international clones (penicillin MIC): S. Africa (19 A)-7 (0.5 microg/ml), Hungary (19 A)-6 (2 microg/ml), Slovakia (19 A)-11 (8 microg/ml), and South Africa (19 A)-13 (8 microg/ml). Fifty-two transformants were characterized by MICs, serogroup-specific PCR, pbp PCR restriction profile and sequence, psp A PCR restriction profile, and erm/mef PCR. A subset was analyzed with multilocus sequence typing (MLST) and pulsed-field gel electrophoresis. Serotype 23 F transformants with penicillin MIC >or= 8 microg/ml were detected through a single transformation with the Hungary (19 A)-6 clone or serial transformations using two to three different clones. Forty-four percent (14/32) of the transformants incorporated >or=1 new MLST allele. Using encapsulated donors, very-high-level penicillin resistant variants of the Tenn (23 F)-4 clone were detected. In addition to detecting stepwise increases in penicillin MIC, a 12-fold increase in penicillin MIC was achieved through a single transformation. This large increase in MIC may explain why this clone is commonly associated with very-high-level resistance in natural populations. Recombination within the MLST housekeeping genes was commonly detected in the transformants that had acquired penicillin resistance.

Predictive models are often constructed from clinical databases with the goal of eventually helping make better clinical decisions. Evaluating models using decision theory is therefore natural. When constructing a model using statistical and machine learning methods, however, we are often uncertain about precisely how the model will be used. Thus, decision-independent measures of classification performance, such as the area under an ROC curve, are popular. As a complementary method of evaluation, we investigate techniques for deriving the expected utility of a model under uncertainty about the model's utilities. We demonstrate an example of the application of this approach to the evaluation of two models that diagnose coronary artery disease.

We investigated two patient-specific and four population-wide machine learning methods for predicting dire outcomes in community acquired pneumonia (CAP) patients. Predicting dire outcomes in CAP patients can significantly influence the decision about whether to admit the patient to the hospital or to treat the patient at home. Population-wide methods induce models that are trained to perform well on average on all future cases. In contrast, patient-specific methods specifically induce a model for a particular patient case. We trained the models on a set of 1601 patient cases and evaluated them on a separate set of 686 cases. One patient-specific method performed better than the population-wide methods when evaluated within a clinically relevant range of the ROC curve. Our study provides support for patient-specific methods being a promising approach for making clinical predictions.

European serotype 14 variants of the France 9V(-3) clone, which have arisen through recombination events involving the penicillin binding protein 1a (pbp1a) gene, have cpsB sequences distinct from those of the 9V(-3) clone. Serotype 14 variants of the 9V(-3) clone have not been compared to genetically diverse serotype 14 strains isolated from an entire metropolitan area in the United States. All serotype 14 non-penicillin-susceptible Streptococcus pneumoniae strains causing invasive disease in Baltimore, Md., from 1995 to 1996 were compared by using pulsed-field gel electrophoresis (PFGE), multilocus sequence typing (MLST), pbp1a PCR restriction profiles, and cpsB and pbp1a sequences. The cpsB genes from strains of 13 serotypes also were analyzed to assess the correlation with serotype. Twenty-seven percent (3 of 11) of the serotype 14 strains were related by PFGE and MLST to the 9V(-3) clone. The serotype 14 variants from Baltimore, unlike the European variants, were related neither to the 9V(-3) clone nor to the R6 strain from positions 1498 to 1710 of the pbp1a gene. All serotype 14 strains had cpsB sequences that differed by or=16% (78 to 83 of 476 bp) divergent from that of the 9V(-3) clone. Allowing for a 2-bp difference in the cpsB sequence resulted in the highest correlation between the cpsB gene and serotype. Overall, 95% (84 of 88) of the strains were classified correctly by serotype with the cpsB sequence. The distal recombination site of the Baltimore serotype 14 variants of the 9V(-3) clone was not identical to that of the European serotype 14 variants. The cpsB gene was serotype specific regardless of whether capsular switching occurred. Although the correlation between serotype and the cpsB sequence was high, the overall diversity of the cpsB gene within a serotype likely will limit the role of this gene in a sequence-based serotyping method.

Detection and prevention of adverse events and, in particular, adverse drug events (ADEs), is an important problem in health care today. We describe the implementation and evaluation of four variations on the simple Bayes model for identifying ADE-related discharge summaries. Our results show that these probabilistic techniques achieve an ROC curve area of up to 0.77 in correctly determining which patient cases should be assigned an ADE-related ICD-9-CM code. These results suggest a potential for these techniques to contribute to the development of an automated system that helps identify ADEs, as a step toward further understanding and preventing them.