Description

Publications in this journal

• Methods to identify standard data elements in clinical and public health forms.

  Authors: Neil F Abernethy, Kathy DeRimer, Peter M Small

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:19-27.

  The fragmentation of clinical and public health systems results in divergent information collection practices, presenting challenges to standardization and EHR certification efforts. Data formsThe fragmentation of clinical and public health systems results in divergent information collection practices, presenting challenges to standardization and EHR certification efforts. Data forms employed in public health jurisdictions nationwide reflect these differences in patient treatment, monitoring and evaluation, and follow-up, presenting challenges for data integration. To study these variations, we surveyed tuberculosis contact investigation forms from all fifty states, three municipalities and two countries. We apply statistics and cluster analysis to analyze the divergent content of contact investigation forms with the goal of characterizing normative practices and identifying a common core of data fields. We found widespread variation in data elements between states in the study, with the "Name" field being the only ubiquitous data element. Our method reveals distinct groupings of data fields employed in certain regions, allowing the simultaneous identification of core standard data fields as well as variations in practice.

• The effect of computerized provider order entry (CPOE) on ordering patterns for chest pain patients in the emergency department.

  Authors: Terrence J Adam, Russ Waitman, Ian Jones, Dominik Aronsky

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:38-47.

  This study addressed the effect of CPOE implementation on chest pain ordering patterns for patients in the emergency department. Retrospective order data was collected to assess the implementation.This study addressed the effect of CPOE implementation on chest pain ordering patterns for patients in the emergency department. Retrospective order data was collected to assess the implementation. 300 randomly selected, time matched patients with a chief complaint of chest pain were selected in a before/after study. Patient demographics, treatment and disposition data were collected on clinical orders. Order volume, completeness and completion times were assessed before and after implementation. Overall order volume increased significantly from 11.6 pre-CPOE to 19.9 post-implementation (p<.01). Order documentation deficiencies were noted pre-implementation with 35.6% containing all order elements. Order completion times were unchanged; however, laboratory completion times increased for admitted patients post-implementation. Order volume increased after CPOE implementation, likely due to improved ED-based admission order capture for admitted patients. Order completeness improved significantly including standing order documentation. Overall, CPOE implementation is associated with improved clinical documentation with limited effect on clinical testing turn-around times.

• Root causes underlying challenges to secondary use of data.

  Authors: Jessica S Ancker, Sarah Shih, Mytri P Singh, Andrew Snyder, Alison Edwards, Rainu Kaushal

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:57-62.

  Although one potential benefit of electronic information systems is the opportunity for secondary use of data, it is often challenging in practice to reuse data. We identify challenges to theAlthough one potential benefit of electronic information systems is the opportunity for secondary use of data, it is often challenging in practice to reuse data. We identify challenges to the secondary use of electronic data from a web-based project management system, and trace these challenges to their root causes. Data quality issues arose from: differential incentives for integrity of different data; software flexibility that allowed a single task to be documented in multiple ways; variability in documentation practices; variability in use of standardized vocabulary; and changes in project procedures and system configuration over time. These issues are very similar to the issues that pose challenges for secondary uses of clinical and operational data for research, public health, and quality improvement. We conclude that secondary use of operational data requires an in-depth understanding of the primary workflow processes that produced the data, as these processes lead to data integrity issues.

• Studying visual behaviors from multiple eye tracking features across levels of information representation.

  Authors: Blake Anderson, Chi-Ren Shyu

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:72-9.

  While eye movements have been used to analyze behaviors for many years, research studies that employ eye tracking technologies are often limited to basic physical features and fixations, which leadsWhile eye movements have been used to analyze behaviors for many years, research studies that employ eye tracking technologies are often limited to basic physical features and fixations, which leads to an abundance of data. Because visual behaviors are complex in nature, it can be difficult to make comparisons and conclusions based on subjects' scanpaths. In this study, we analyze visual activities from 15 expert radiographers and 26 novices as they view a series of images to attempt to discover relationships between a large number of features including fixation, region and subject information. We expect that the techniques used in this study will be useful in finding common behaviors in eye tracking data for medical applications. These behaviors could be used to train novices and prevent potential medical errors that occur during visual analysis of medical images.

• Technical architecture of ONC-approved plans for statewide health information exchange.

  Authors: Randolph C Barrows, John Ezzard

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:88-97.

  ONC-approved state plans for HIE were reviewed for descriptions and depictions of statewide HIE technical architecture. Review was complicated by non-standard organizational elements and technicalONC-approved state plans for HIE were reviewed for descriptions and depictions of statewide HIE technical architecture. Review was complicated by non-standard organizational elements and technical terminology across state plans. Findings were mapped to industry standard, referenced, and defined HIE architecture descriptions and characteristics. Results are preliminary due to the initial subset of ONC-approved plans available, the rapid pace of new ONC-plan approvals, and continuing advancements in standards and technology of HIE, etc. Review of 28 state plans shows virtually all include a direct messaging component, but for participating entities at state-specific levels of granularity (RHIO, enterprise, organization/provider). About ½ of reviewed plans describe a federated architecture, and ¼ of plans utilize a single-vendor "hybrid-federated" architecture. About 1/3 of states plan to leverage new federal and open exchange technologies (DIRECT, CONNECT, etc.). Only one plan describes a centralized architecture for statewide HIE, but others combine central and federated architectural approaches.

• Automated plan-recognition of chemotherapy protocols.

  Authors: Haresh Bhatia, Mia Levy

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:108-14.

  Cancer patients are often treated with multiple sequential chemotherapy protocols ranging in complexity from simple to highly complex patterns of multiple repeating drugs. Clinical documentationCancer patients are often treated with multiple sequential chemotherapy protocols ranging in complexity from simple to highly complex patterns of multiple repeating drugs. Clinical documentation procedures that focus on details of single drug events, however, make it difficult for providers and systems to efficiently abstract the sequence and nature of treatment protocols. We have developed a data driven method for cancer treatment plan recognition that takes as input pharmacy chemotherapy dispensing records and produces the sequence of identified chemotherapy protocols. Compared to a manually annotated gold standard, our method was 75% accurate and 80% precise for a breast cancer testing set (110 patients, 2,029 drug events), and 54% accurate and 63% precise for a lung cancer testing set (53 patients, 670 drug events). This method for cancer treatment plan recognition may provide clinicians and systems an abstracted view of the patient's treatment history.

• Revisiting the EBM decision model to formalize non-compliance with computerized CPGs: results in the management of breast cancer with OncoDoc2.

  Authors: Jacques Bouaud, Brigitte Séroussi

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:125-34.

  In 2002, Haynes et al. founded a prescriptive model of evidence-based medicine based on the patient's clinical state, her preferences, and research evidence, clinical expertise synthesizing the otherIn 2002, Haynes et al. founded a prescriptive model of evidence-based medicine based on the patient's clinical state, her preferences, and research evidence, clinical expertise synthesizing the other three components. Revisiting this model of medical decision making, we propose a descriptive model introducing clinicians' preferences and formalize four reasons of non-compliance with clinical practice guidelines (CPGs). The approach has been applied to breast cancer management decisions taken by multidisciplinary staff meetings (MSMs) at the Tenon hospital, Paris, France, while using a clinical decision support system (CDSS): OncoDoc2. 1,889 MSM decisions have been recorded [February 2007-October 2009]. The compliance rate with CPGs was measured at 91.0%. Non-compliant decisions are mainly "MSM choices" (39.1%) and "particular cases" (34.9%). "Practice evolution" and "patient choices" are less frequent (12.4% and 11.2%). Even with a CDSS, a 100% compliance rate cannot be reached because particular cases fall outside CPGs and borderline cases need to be interpreted by clinicians.

• Progress and challenge in meeting meaningful use at an integrated delivery network.

  Authors: Watson A Bowes

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:144-51.

  Intermountain Healthcare hospitals and providers are eligible for approximately $95 million in incentives from the Health Information Technology for Economic and Clinical Health Act (HITECH), whichIntermountain Healthcare hospitals and providers are eligible for approximately $95 million in incentives from the Health Information Technology for Economic and Clinical Health Act (HITECH), which requires that hospitals and providers use a certified electronic health record (EHR) in a meaningful way. This paper describes the our progress in readying legacy systems for certification, including measuring, and filling gaps in (EHR) functionality. Also addressed are some of the challenges and successes in meeting meaningful use. Methods for measuring and tracking levels of clinician meaningful use behaviors, and our most recent results impacting meaningful use behaviors in a large integrated delivery network are described. We identified 20 EHR requirements we can certify now, 16 requirements with minor issues to resolve, and 38 requirements which are still in some state of development. We also identified 6 meaningful use workflows that will require significant work to bring all of our hospitals and providers above the measure requirement.

• Simulation Analysis Platform (SnAP): a tool for evaluation of public health surveillance and disease control strategies.

  Authors: David L Buckeridge, Christian Jauvin, Anya Okhmatovskaia, Aman D Verma

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:161-70.

  Increasingly, researchers use simulation to generate realistic population health data to evaluate surveillance and disease control methods. This evaluation approach is attractive because real dataIncreasingly, researchers use simulation to generate realistic population health data to evaluate surveillance and disease control methods. This evaluation approach is attractive because real data are often not available to describe the full range of population health trajectories that may occur. Simulation models, especially agent-based models, tend to have many parameters and it is often difficult for researchers to evaluate the effect of the multiple parameter values on model outcomes. In this paper, we describe Simulation Analysis Platform (SnAP) - a software infrastructure for automatically deploying and analyzing multiple runs of a simulation model in a manner that efficiently explores the influence of parameter uncertainty and random error on model outcomes. SnAP is designed to be efficient, scalable, extensible, and portable. We describe the design decisions taken to meet these requirements, present the design of the platform, and describe results from an example application of SnAP.

• Can SNOMED CT fulfill the vision of a compositional terminology? Analyzing the use case for problem list.

  Authors: James R Campbell, Junchuan Xu, Kin Wah Fung

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:181-8.

  We analyzed 598 of 63,952 terms employed in problem list entries from seven major healthcare institutions that were not mapped with UMLS to SNOMED CT when preparing the NLM UMLS-CORE problem listWe analyzed 598 of 63,952 terms employed in problem list entries from seven major healthcare institutions that were not mapped with UMLS to SNOMED CT when preparing the NLM UMLS-CORE problem list subset. We intended to determine whether published or post-coordinated SNOMED concepts could accurately capture the problems as stated by the clinician and to characterize the workload for the local terminology manager. From the terms we analyzed, we estimate that 7.5% of the total terms represent ambiguous statements that require clarification. Of those terms which were unambiguous, we estimate that 38.1% could be encoded using the SNOMED CT January 2011 pre-coordinated (published core) content. 60.4% of unambiguous terms required post-coordination to capture the term meaning within the SNOMED model. Approximately 28.5% of post-coordinated content could not be fully defined and required primitive forms. This left 1.5% of unambiguous terms which were expressed with meaning which could not be represented in SNOMED CT. We estimate from our study that 98.5% of clinical terms unambiguously suggested for the problem list can be equated to published concepts or can be modeled with SNOMED CT but that roughly one in four SNOMED modeled expressions fail to represent the full meaning of the term. Implications for the business model of the local terminology manager and the development of SNOMED CT are discussed.

• SNOMED CT revisions and coded data repositories: when to upgrade?

  Authors: Werner Ceusters

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:197-206.

  SNOMED CT is gaining momentum in its acceptance and operational application as a reference terminology in electronic health systems. Because it is revised every six months, organizations using SNOMEDSNOMED CT is gaining momentum in its acceptance and operational application as a reference terminology in electronic health systems. Because it is revised every six months, organizations using SNOMED CT might feel a need to ensure that their systems are synchronized with these revisions. It has been shown that for certain sorts of applications migration to a new version is a labor-intensive process. Here two indicators - the evolution of the global information content of an ontology over consecutive versions, and the perseverance of suspicious events - are proposed to assess whether it is worthwhile upgrading when a new version is released. The indicators can be computed automatically when a new version is released and are statistically unrelated. Trend breaks in their evolution are suggestive for the possible benefit of an upgrade and their predictive power correlates well with the retrospective realism-based quality metric which forms the basis of Evolutionary Terminology Auditing.

• Predicting adverse drug events from personal health messages.

  Authors: Brant W Chee, Richard Berlin, Bruce Schatz

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:217-26.

  Adverse drug events (ADEs) remain a large problem in the United States, being the fourth leading cause of death, despite post market drug surveillance. Much post consumer drug surveillance relies onAdverse drug events (ADEs) remain a large problem in the United States, being the fourth leading cause of death, despite post market drug surveillance. Much post consumer drug surveillance relies on self-reported "spontaneous" patient data. Previous work has performed datamining over the FDA's Adverse Event Reporting System (AERS) and other spontaneous reporting systems to identify drug interactions and drugs correlated with high rates of serious adverse events. However, safety problems have resulted from the lack of post marketing surveillance information about drugs, with underreporting rates of up to 98% within such systems. We explore the use of online health forums as a source of data to identify drugs for further FDA scrutiny. In this work we aggregate individuals' opinions and review of drugs similar to crowd intelligence3. We use natural language processing to group drugs discussed in similar ways and are able to successfully identify drugs withdrawn from the market based on messages discussing them before their removal.

• Achieving reliable communication in dynamic emergency responses.

  Authors: Octav Chipara, Anders N Plymoth, Fang Liu, Ricky Huang, Brian Evans, Per Johansson, Ramesh Rao, William G Griswold

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:238-47.

  Emergency responses require the coordination of first responders to assess the condition of victims, stabilize their condition, and transport them to hospitals based on the severity of theirEmergency responses require the coordination of first responders to assess the condition of victims, stabilize their condition, and transport them to hospitals based on the severity of their injuries. WIISARD is a system designed to facilitate the collection of medical information and its reliable dissemination during emergency responses. A key challenge in WIISARD is to deliver data with high reliability as first responders move and operate in a dynamic radio environment fraught with frequent network disconnections. The initial WIISARD system employed a client-server architecture and an ad-hoc routing protocol was used to exchange data. The system had low reliability when deployed during emergency drills. In this paper, we identify the underlying causes of unreliability and propose a novel peer-to-peer architecture that in combination with a gossip-based communication protocol achieves high reliability. Empirical studies show that compared to the initial WIISARD system, the redesigned system improves reliability by as much as 37% while reducing the number of transmitted packets by 23%.

• Improving adherence to research protocol drug exclusions using a clinical alerting system.

  Authors: James J Cimino, Lincoln Farnum, Gary E DiPatrizio, Barry R Goldspiel

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:257-66.

  To develop a general method for using the alerting function of an electronic health record (EHR) system to warn prescribers when a drug order may be in conflict with the restrictions of a patient'sTo develop a general method for using the alerting function of an electronic health record (EHR) system to warn prescribers when a drug order may be in conflict with the restrictions of a patient's research protocol. We examined a sample of clinical research protocols at the National Institutes of Health (NIH) to identify the frequency with which drugs were excluded by protocols. We analyzed two protocols and modeled the exclusions they contained. We then developed a data model to represent the exclusions, expanded the terminology in the NIH's Biomedical Translational Research Information System (BTRIS) to include relevant drug concepts, and wrote a medical logic module (MLM) for the EHR to match terms for ordered drugs with the drug concepts in the protocol. We found that 50% of protocols in our sample included drug exclusions. Our model represented exclusion concepts and also concepts related to exemptions from the exclusions. The MLM was deployed in a test environment where it successfully detected orders for excluded drugs and delivered messages to users explaining the exclusion, providing information about the clinical setting and timing where the exclusion applies. BTRIS reports using the same terminology information were able to identify instances where protocol exceptions occurred. Drug exclusions are frequent components of research protocols; nonadherenece to these exclusions could result in harm to subjects, erroneous study results or inefficiencies due to disqualification of research subjects. Our approach uses an MLM and a simple knowledge base, together with a controlled terminology, to provide a solution to the detection and prevention of possible protocol violations. Further work is needed to model additional aspects of the exclusions, such as timing and co-occurring conditions, to improve MLM accuracy.

• Analyzing the heterogeneity and complexity of Electronic Health Record oriented phenotyping algorithms.

  Authors: Mike Conway, Richard L Berg, David Carrell, Joshua C Denny, Abel N Kho, Iftikhar J Kullo, James G Linneman, Jennifer A Pacheco, Peggy Peissig, Luke Rasmussen, Noah Weston, Christopher G Chute, Jyotishman Pathak

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:274-83.

  The need for formal representations of eligibility criteria for clinical trials - and for phenotyping more generally - has been recognized for some time. Indeed, the availability of a formalThe need for formal representations of eligibility criteria for clinical trials - and for phenotyping more generally - has been recognized for some time. Indeed, the availability of a formal computable representation that adequately reflects the types of data and logic evidenced in trial designs is a prerequisite for the automatic identification of study-eligible patients from Electronic Health Records. As part of the wider process of representation development, this paper reports on an analysis of fourteen Electronic Health Record oriented phenotyping algorithms (developed as part of the eMERGE project) in terms of their constituent data elements, types of logic used and temporal characteristics. We discovered that the majority of eMERGE algorithms analyzed include complex, nested boolean logic and negation, with several dependent on cardinality constraints and complex temporal logic. Insights gained from the study will be used to augment the CDISC Protocol Representation Model.

• Medical record and imaging evaluation to identify arterial tortuosity phenotype in populations at risk for intracranial aneurysms.

  Authors: Karl T Diedrich, John A Roberts, Richard H Schmidt, Lisa A Cannon Albright, Anji T Yetman, Dennis L Parker

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:295-304.

  High arterial tortuosity may signify early arterial pathology which may precede development of intracranial aneurysms. We measured arterial tortuosity of intracranial vessels and reviewed the medicalHigh arterial tortuosity may signify early arterial pathology which may precede development of intracranial aneurysms. We measured arterial tortuosity of intracranial vessels and reviewed the medical records of three groups of patients: with intracranial aneurysms, without aneurysms but at increased clinical risk, and controls without aneurysms or associated risk factors. There was significant but inconsistent evidence of increased arterial tortuosity in aneurysm cases and high-risk cases across different arteries. Medical records review identified that a subset of aneurysm cases carried a diagnosis of Loeys-Dietz syndrome that is often misdiagnosed as Marfan syndrome. We found increased arterial tortuosity in the Loeys-Dietz syndrome cases. A combination of medical record screening for Marfan syndrome or Loeys-Dietz symptoms such as aneurysms and evaluation of arterial tortuosity by a curve of scores from medical images may identify previously undiagnosed cases of Loeys-Dietz syndrome.

• A generative model based approach to retrieving ischemic stroke images.

  Authors: Thien Anh Dinh, Tomi Silander, C C Tchoyoson Lim, Tze-Yun Leong

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:312-21.

  This paper proposes a generative model approach to automatically annotate medical images to improve the efficiency and effectiveness of image retrieval systems for teaching, research, and diagnosis.This paper proposes a generative model approach to automatically annotate medical images to improve the efficiency and effectiveness of image retrieval systems for teaching, research, and diagnosis. The generative model captures the probabilistic relationships among relevant classification tags, tentative lesion patterns, and selected input features. Operating on the imperfect segmentation results of input images, the probabilistic framework can effectively handle the inherent uncertainties in the images and insufficient information in the training data. Preliminary assessment in the ischemic stroke subtype classification shows that the proposed system is capable of generating the relevant tags for ischemic stroke brain images. The main benefit of this approach is its scalability; the method can be applied in large image databases as it requires only minimal manual labeling of the training data and does not demand high-precision segmentation of the images.

• From simply inaccurate to complex and inaccurate: complexity in standards-based quality measures.

  Authors: David A Dorr, Aaron M Cohen, Marsha Pierre-Jacques Williams, John Hurdle

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:331-8.

  Quality measurement has been slow to make a major impact in health care. Initial measures were too simple to affect outcomes of importance. Incentive programs such as Meaningful Use encourage betterQuality measurement has been slow to make a major impact in health care. Initial measures were too simple to affect outcomes of importance. Incentive programs such as Meaningful Use encourage better measures, but in process may become more complex. We evaluated the measures selected for Meaningful Use in two ways: we counted unique concept identifiers, taxonomies, and aggregated concepts as measures of complexity; and we surveyed informatics professionals to assess difficulty. There were 20,316 unique concept identifiers, 35 taxonomies, and 317 aggregated concepts across the 45 measures. Half the respondents reported measures at least moderately difficult. The number of identifiers was associated with fewer implementations (r=-.37); rating-of-difficulty was associated with more taxonomies (r=.24). The impact on accuracy may be substantial when moving to measures intended to be more relevant to clinical outcomes but requiring the use of more taxonomies, unused structured concept identifiers, or concepts only in free text fields.

• Integrating machine learning and physician knowledge to improve the accuracy of breast biopsy.

  Authors: I Dutra, H Nassif, D Page, J Shavlik, R M Strigel, Y Wu, M E Elezaby, E Burnside

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:349-55.

  In this work we show that combining physician rules and machine learned rules may improve the performance of a classifier that predicts whether a breast cancer is missed on percutaneous, image-guidedIn this work we show that combining physician rules and machine learned rules may improve the performance of a classifier that predicts whether a breast cancer is missed on percutaneous, image-guided breast core needle biopsy (subsequently referred to as "breast core biopsy"). Specifically, we show how advice in the form of logical rules, derived by a sub-specialty, i.e. fellowship trained breast radiologists (subsequently referred to as "our physicians") can guide the search in an inductive logic programming system, and improve the performance of a learned classifier. Our dataset of 890 consecutive benign breast core biopsy results along with corresponding mammographic findings contains 94 cases that were deemed non-definitive by a multidisciplinary panel of physicians, from which 15 were upgraded to malignant disease at surgery. Our goal is to predict upgrade prospectively and avoid surgery in women who do not have breast cancer. Our results, some of which trended toward significance, show evidence that inductive logic programming may produce better results for this task than traditional propositional algorithms with default parameters. Moreover, we show that adding knowledge from our physicians into the learning process may improve the performance of the learned classifier trained only on data.

• A cloud-based simulation architecture for pandemic influenza simulation.

  Authors: Henrik Eriksson, Massimiliano Raciti, Maurizio Basile, Alessandro Cunsolo, Anders Fröberg, Ola Leifler, Joakim Ekberg, Toomas Timpka

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:364-73.

  High-fidelity simulations of pandemic outbreaks are resource consuming. Cluster-based solutions have been suggested for executing such complex computations. We present a cloud-based simulationHigh-fidelity simulations of pandemic outbreaks are resource consuming. Cluster-based solutions have been suggested for executing such complex computations. We present a cloud-based simulation architecture that utilizes computing resources both locally available and dynamically rented online. The approach uses the Condor framework for job distribution and management of the Amazon Elastic Computing Cloud (EC2) as well as local resources. The architecture has a web-based user interface that allows users to monitor and control simulation execution. In a benchmark test, the best cost-adjusted performance was recorded for the EC2 H-CPU Medium instance, while a field trial showed that the job configuration had significant influence on the execution time and that the network capacity of the master node could become a bottleneck. We conclude that it is possible to develop a scalable simulation environment that uses cloud-based solutions, while providing an easy-to-use graphical user interface.

• Part-of-speech tagging for clinical text: wall or bridge between institutions?

  Authors: Jung-wei Fan, Rashmi Prasad, Rommel M Yabut, Richard M Loomis, Daniel S Zisook, John E Mattison, Yang Huang

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:382-91.

  Part-of-speech (POS) tagging is a fundamental step required by various NLP systems. The training of a POS tagger relies on sufficient quality annotations. However, the annotation process is bothPart-of-speech (POS) tagging is a fundamental step required by various NLP systems. The training of a POS tagger relies on sufficient quality annotations. However, the annotation process is both knowledge-intensive and time-consuming in the clinical domain. A promising solution appears to be for institutions to share their annotation efforts, and yet there is little research on associated issues. We performed experiments to understand how POS tagging performance would be affected by using a pre-trained tagger versus raw training data across different institutions. We manually annotated a set of clinical notes at Kaiser Permanente Southern California (KPSC) and a set from the University of Pittsburg Medical Center (UPMC), and trained/tested POS taggers with intra- and inter-institution settings. The cTAKES POS tagger was also included in the comparison to represent a tagger partially trained from the notes of a third institution, Mayo Clinic at Rochester. Intra-institution 5-fold cross-validation estimated an accuracy of 0.953 and 0.945 on the KPSC and UPMC notes respectively. Trained purely on KPSC notes, the accuracy was 0.897 when tested on UPMC notes. Trained purely on UPMC notes, the accuracy was 0.904 when tested on KPSC notes. Applying the cTAKES tagger pre-trained with Mayo Clinic's notes, the accuracy was 0.881 on KPSC notes and 0.883 on UPMC notes. After adding UPMC annotations to KPSC training data, the average accuracy on tested KPSC notes increased to 0.965. After adding KPSC annotations to UPMC training data, the average accuracy on tested UPMC notes increased to 0.953. The results indicated: first, the performance of pre-trained POS taggers dropped about 5% when applied directly across the institutions; second, mixing annotations from another institution following the same guideline increased tagging accuracy for about 1%. Our findings suggest that institutions can benefit more from sharing raw annotations but less from sharing pre-trained models for the POS tagging task. We believe the study could also provide general insights on cross-institution data sharing for other types of NLP tasks.

• Practical challenges in the secondary use of real-world data: the notifiable condition detector.

  Authors: Mustafa Fidahussein, Jeff Friedlin, Shaun Grannis

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:402-8.

  The interoperability specifications for electronic laboratory reporting specify the use of HL7, LOINC, SNOMED CT and UCUM. We explored the degree to which health care transactions comply with theseThe interoperability specifications for electronic laboratory reporting specify the use of HL7, LOINC, SNOMED CT and UCUM. We explored the degree to which health care transactions comply with these standards by evaluating laboratory data captured in a health information exchange to support automated detection of public health notifiable diseases. We studied the NCD's ability to detect and report Lead, Influenza and MRSA. We found that due to incomplete LOINC mapping, alternate approaches such as keyword searches within local test names and codes could identify additional potentially reportable messages. We also found that non-adherence to HL7 messaging standards and inconsistently recorded laboratory results require the use of complex systems with complementary NLP techniques to accurately report notifiable conditions. We conclude that the incomplete adoption of and adherence to specified standards poses challenges to deploying processes that utilize real-world data for secondary purposes.

• Workflow concerns and workarounds of readers in an urban safety net teleretinal screening study.

  Authors: Allison Fish, Sheba George, Elizabeth Terrien, Alicia Eccles, Richard Baker, Omolola Ogunyemi

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:417-26.

  Telemedicine holds great promise for increased access to specialty care services for safety net clinic patients. However, the adoption of these technologies is not a seamless transition forTelemedicine holds great promise for increased access to specialty care services for safety net clinic patients. However, the adoption of these technologies is not a seamless transition for clinicians working in resource-poor settings. Previous research has analyzed workflow issues that arise in primary care settings when adopting telehealth tools but has not examined the unique workflow challenges facing specialists who provide assessments to safety net clinics. Findings are presented from a case study that employed qualitative methodologies as part of an assessment of a teleretinal screening program in Los Angeles urban safety net clinics. The program utilizes external ophthalmologists to perform retinal readings. The case study provides insights into how difficulties that arise in reader workflow are resolved and identifies unique factors requiring consideration when highly trained specialists perform teleretinal readings. The discussion outlines important issues to address when developing telehealth workflow protocols for the safety net, specifically, and their broader applicability in telemedicine.

• An evaluation of the UMLS in representing corpus derived clinical concepts.

  Authors: Jeff Friedlin, Marc Overhage

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:435-44.

  We performed an evaluation of the Unified Medical Language System (UMLS) in representing concepts derived from medical narrative documents from three domains: chest x-ray reports, discharge summariesWe performed an evaluation of the Unified Medical Language System (UMLS) in representing concepts derived from medical narrative documents from three domains: chest x-ray reports, discharge summaries and admission notes. We detected concepts in these documents by identifying noun phrases (NPs) and N-grams, including unigrams (single words), bigrams (word pairs) and trigrams (word triples). After removing NPs and N-grams that did not represent discrete clinical concepts, we processed the remaining with the UMLS MetaMap program. We manually reviewed the results of MetaMap processing to determine whether MetaMap found full, partial or no representation of the concept. For full representations, we determined whether post-coordination was required. Our results showed that a large portion of concepts found in clinical narrative documents are either unrepresented or poorly represented in the current version of the UMLS Metathesaurus and that post-coordination was often required in order to fully represent a concept.

• Understanding the work of pediatric inpatient medicine teams: implications for information system requirements.

  Authors: Ching-Ping Lin, John H Gennari

  AMIA ... Annual Symposium proceedings / AMIA Symposium. AMIA Symposium. 2011:455-64.

  Health information systems tend to be designed primarily for data retrieval and data entry, with insufficient attention paid to the larger contexts in which work occurs. As a result, low physicianHealth information systems tend to be designed primarily for data retrieval and data entry, with insufficient attention paid to the larger contexts in which work occurs. As a result, low physician acceptance and satisfaction remain barriers to the successful integration of current informational and decision support systems (e.g., CPOE systems). This paper reports on a qualitative field study of team-based clinical care work and decision making. Our aim is to consider a radical redesign of clinical information systems, one that is built with context and the constraints of work practice in mind. We apply Cognitive Work Analysis (CWA) methods to characterize the work at a large children's hospital, and analyze these results to consider new designs for clinical information systems. In this study, we report on themes, constraints, and ideas for design, showing how our CWA analyses lead to designs that are very different from current technology information systems.

Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.