Henry J. Lowe

STRIDE--An integrated standards-based translational research informatics platform.

STRIDE (Stanford Translational Research Integrated Database Environment) is a research and development project at Stanford University to create a standards-based informatics platform supporting clinical and translational research. STRIDE consists of three integrated components: a clinical data warehouse, based on the HL7 Reference Information Model (RIM), containing clinical information on over 1.3 million pediatric and adult patients cared for at Stanford University Medical Center since 1995; an application development framework for building research data management applications on the STRIDE platform and a biospecimen data management system. STRIDEs semantic model uses standardized terminologies, such as SNOMED, RxNorm, ICD and CPT, to represent important biomedical concepts and their relationships. The system is in daily use at Stanford and is an important component of Stanford Universitys CTSA (Clinical and Translational Science Award) Informatics Program.on behalf of the American Heart Association Statistics Committee and Stroke Statistics Nathan D. Wong, Daniel Woo and Melanie B. Turner Elsayed Z. Soliman, Paul D. Sorlie, Nona Sotoodehnia, Tanya N. Turan, Salim S. Virani, Claudia S. Moy, Dariush Mozaffarian, Michael E. Mussolino, Graham Nichol, Nina P. Paynter, Lynda D. Lisabeth, Diane M. Makuc, Gregory M. Marcus, Ariane Marelli, David B. Matchar, Lichtman, Virginia J. Howard, Brett M. Kissela, Steven J. Kittner, Daniel T. Lackland, Judith H. Caroline S. Fox, Heather J. Fullerton, Cathleen Gillespie, Susan M. Hailpern, John A. Heit, Benjamin, Jarett D. Berry, William B. Borden, Dawn M. Bravata, Shifan Dai, Earl S. Ford, Writing Group Members, Véronique L. Roger, Alan S. Go, Donald M. Lloyd-Jones, Emelia J. Association 2012 Update : A Report From the American Heart −− Heart Disease and Stroke StatisticsHeart Disease, Stroke and other Cardiovascular Diseases • Cardiovascular disease is the leading global cause of death, accounting for 17.3 million deaths per year, a number that is expected to grow to more than 23.6 million by 2030. • In 2008, cardiovascular deaths represented 30 percent of all global deaths, with 80 percent of those deaths taking place in lowand middle-income countries. • Nearly 787,000 people in the U.S. died from heart disease, stroke and other cardiovascular diseases in 2011. That’s about one of every three deaths in America. • About 2,150 Americans die each day from these diseases, one every 40 seconds. • Cardiovascular diseases claim more lives than all forms of cancer combined. • About 85.6 million Americans are living with some form of cardiovascular disease or the after-effects of stroke. • Direct and indirect costs of cardiovascular diseases and stroke total more than

A Novel Hybrid Approach to Automated Negation Detection in Clinical Radiology Reports

320.1 billion. That includes health expenditures and lost productivity. • Nearly half of all African-American adults have some form of cardiovascular disease, 48 percent of women and 46 percent of men. • Heart disease is the No. 1 cause of death in the world and the leading cause of death in the United States, killing over 375,000 Americans a year. • Heart disease accounts for 1 in 7 deaths in the U.S. • Someone in the U.S. dies from heart disease about once every 90 seconds.Author(s): Mozaffarian, Dariush; Benjamin, Emelia J; Go, Alan S; Arnett, Donna K; Blaha, Michael J; Cushman, Mary; de Ferranti, Sarah; Despres, Jean-Pierre; Fullerton, Heather J; Howard, Virginia J; Huffman, Mark D; Judd, Suzanne E; Kissela, Brett M; Lackland, Daniel T; Lichtman, Judith H; Lisabeth, Lynda D; Liu, Simin; Mackey, Rachel H; Matchar, David B; McGuire, Darren K; Mohler, Emile R; Moy, Claudia S; Muntner, Paul; Mussolino, Michael E; Nasir, Khurram; Neumar, Robert W; Nichol, Graham; Palaniappan, Latha; Pandey, Dilip K; Reeves, Mathew J; Rodriguez, Carlos J; Sorlie, Paul D; Stein, Joel; Towfighi, Amytis; Turan, Tanya N; Virani, Salim S; Willey, Joshua Z; Woo, Daniel; Yeh, Robert W; Turner, Melanie B; American Heart Association Statistics Committee and Stroke Statistics Subcommittee

A Randomized Trial Using Computerized Decision Support to Improve Treatment of Major Depression in Primary Care

OBJECTIVE Negation is common in clinical documents and is an important source of poor precision in automated indexing systems. Previous research has shown that negated terms may be difficult to identify if the words implying negations (negation signals) are more than a few words away from them. We describe a novel hybrid approach, combining regular expression matching with grammatical parsing, to address the above limitation in automatically detecting negations in clinical radiology reports. DESIGN Negations are classified based upon the syntactical categories of negation signals, and negation patterns, using regular expression matching. Negated terms are then located in parse trees using corresponding negation grammar. MEASUREMENTS A classification of negations and their corresponding syntactical and lexical patterns were developed through manual inspection of 30 radiology reports and validated on a set of 470 radiology reports. Another 120 radiology reports were randomly selected as the test set on which a modified Delphi design was used by four physicians to construct the gold standard. RESULTS In the test set of 120 reports, there were a total of 2,976 noun phrases, of which 287 were correctly identified as negated (true positives), along with 23 undetected true negations (false negatives) and 4 mistaken negations (false positives). The hybrid approach identified negated phrases with sensitivity of 92.6% (95% CI 90.9-93.4%), positive predictive value of 98.6% (95% CI 96.9-99.4%), and specificity of 99.87% (95% CI 99.7-99.9%). CONCLUSION This novel hybrid approach can accurately locate negated concepts in clinical radiology reports not only when in close proximity to, but also at a distance from, negation signals.

The World Wide Web: A Review of an Emerging Internet-based Technology for the Distribution of Biomedical Information

OBJECTIVE: To examine whether feedback and treatment advice for depression presented to primary care physicians (PCPs) via an electronic medical record (EMR) system can potentially improve clinical outcomes and care processes for patients with major depression.DESIGN: Randomized controlled trial.SETTING: Academically affiliated primary care practice in Pittsburgh, PA.PATIENTS: Two hundred primary care patients with major depression on the Primary Care Evaluation of Mental Disorders (PRIME-MD) and who met all protocol-eligibility criteria.INTERVENTION: PCPs were randomly assigned to 1 of 3 levels of exposure to EMR feedback of guideline-based treatment advice for depression: “active care” (AC), “passive care” (PC), or “usual care” (UC).MEASUREMENTS AND MAIN RESULTS: Patients’ 3- and 6-month Hamilton Rating Scale for Depression (HRS-D) score and chart review of PCP reports of depression care in the 6 months following the depression diagnosis. Only 22% of patients recovered from their depressive episode at 6 months (HRS-D ≤7). Patients’ mean HRS-D score decreased regardless of their PCPs’ guideline-exposure condition (20.4 to 14.2 from baseline to 6-month follow-up; P<.001). However, neither continuous (HRS-D ≤7: 22% AC, 23% PC, 22% UC; P=.8) nor categorical measures of recovery (P=.2) differed by EMR exposure condition upon follow-up. Care processes for depression were also similar by PCP assignment despite exposure to repeated reminders of the depression diagnosis and treatment advice (e.g., depression mentioned in ≥3 contacts with usual PCP at 6 months: 31% AC, 31% PC, 18% UC; P=.09 and antidepressant medication suggested/prescribed or baseline regimen modified at 6 months: 59% AC, 57% PC, 52% UC; P=.3).CONCLUSIONS: Screening for major depression, electronically informing PCPs of the diagnosis, and then exposing them to evidence-based treatment recommendations for depression via EMR has little differential impact on patients’ 3- or 6-month clinical outcomes or on process measures consistent with high-quality depression care.

Improved Identification of Noun Phrases in Clinical Radiology Reports Using a High-Performance Statistical Natural Language Parser Augmented with the UMLS Specialist Lexicon

The Internet is rapidly evolving from a resource used primarily by the research community to a true global information network offering a wide range of databases and services. This evolution presents many opportunities for improved access to biomedical information, but Internet-based resources have often been difficult for the non-expert to develop and use. The World Wide Web (WWW) supports an inexpensive, easy-to-use, cross-platform, graphic interface to the Internet that may radically alter the way we retrieve and disseminate medical data. This paper summarizes the Internet and hypertext origins of the WWW, reviews WWW-specific technologies, and describes current and future applications of this technology in medicine and medical informatics. The paper also includes an appendix of useful biomedical WWW servers.

Selective Automated Indexing of Findings and Diagnoses in Radiology Reports

OBJECTIVE The aim of this study was to develop and evaluate a method of extracting noun phrases with full phrase structures from a set of clinical radiology reports using natural language processing (NLP) and to investigate the effects of using the UMLS(R) Specialist Lexicon to improve noun phrase identification within clinical radiology documents. DESIGN The noun phrase identification (NPI) module is composed of a sentence boundary detector, a statistical natural language parser trained on a nonmedical domain, and a noun phrase (NP) tagger. The NPI module processed a set of 100 XML-represented clinical radiology reports in Health Level 7 (HL7)(R) Clinical Document Architecture (CDA)-compatible format. Computed output was compared with manual markups made by four physicians and one author for maximal (longest) NP and those made by one author for base (simple) NP, respectively. An extended lexicon of biomedical terms was created from the UMLS Specialist Lexicon and used to improve NPI performance. RESULTS The test set was 50 randomly selected reports. The sentence boundary detector achieved 99.0% precision and 98.6% recall. The overall maximal NPI precision and recall were 78.9% and 81.5% before using the UMLS Specialist Lexicon and 82.1% and 84.6% after. The overall base NPI precision and recall were 88.2% and 86.8% before using the UMLS Specialist Lexicon and 93.1% and 92.6% after, reducing false-positives by 31.1% and false-negatives by 34.3%. CONCLUSION The sentence boundary detector performs excellently. After the adaptation using the UMLS Specialist Lexicon, the statistical parsers NPI performance on radiology reports increased to levels comparable to the parsers native performance in its newswire training domain and to that reported by other researchers in the general nonmedical domain.

Computerized Physician Order Entry in the Critical Care Environment: A Review of Current Literature

The recent improvements in capabilities of desktop computers and communications networks give impetus for the development of clinical image repositories that can be used for patient care and medical education. A challenge in the use of these systems is the accurate indexing of images for retrieval performance acceptable to users. This paper describes a series of experiments aiming to adapt the SAPHIRE system, which matches text to concepts in the UMLS Metathesaurus, for the automated indexing of image reports. A series of enhancements to the baseline system resulted in a recall of 63% but a precision of only 30% in detecting concepts. At this level of performance, such a system might be problematic for users in a purely automated indexing environment. However, if the ability to retrieve images in repositories based on content in their reports is desired by clinical users, and no other current systems offer this functionality, then follow-up research questions include whether these imperfect results would be useful in a completely or partially automated indexing environment and/or whether other approaches can improve upon them.

A simple heuristic for blindfolded record linkage.

The implementation of health information technology (HIT) is accelerating, driven in part by a growing interest in computerized physician order entry (CPOE) as a tool for improving the quality and safety of patient care. Computerized physician order entry could have a substantial impact on patients in intensive care, where the potential for medical error is high, and the clinical workflow is complex. In 2009, only 17% of hospitals had functional CPOE systems in place. In intensive care unit (ICU) settings, CPOE has been shown to reduce the occurrence of some medication errors, but evidence of a beneficial effect on clinical outcomes remains limited. In some cases, new error types have arisen with the use of CPOE. Intensive care unit workflow and staff relationships have been affected by CPOE, often in unanticipated ways. The design of CPOE software has a strong impact on user acceptance. Intensive care unit-specific order sets lessen the cognitive workload associated with the use of CPOE and improve user acceptance. The diffusion of new technological innovations in the ICU can have unintended consequences, including changes in workflow, staff roles, and patient outcomes. When implementing CPOE in critical care areas, both organizational and technical factors should be considered. Further research is needed to inform the design and management of CPOE systems in the ICU and to better assess their impact on clinical end points, cost-effectiveness, and user satisfaction.

Discretization of continuous features in clinical datasets

OBJECTIVES To address the challenge of balancing privacy with the need to create cross-site research registry records on individual patients, while matching the data for a given patient as he or she moves between participating sites. To evaluate the strategy of generating anonymous identifiers based on real identifiers in such a way that the chances of a shared patient being accurately identified were maximized, and the chances of incorrectly joining two records belonging to different people were minimized. METHODS Our hypothesis was that most variation in names occurs after the first two letters, and that date of birth is highly reliable, so a single match variable consisting of a hashed string built from the first two letters of the patients first and last names plus their date of birth would have the desired characteristics. We compared and contrasted the match algorithm characteristics (rate of false positive v. rate of false negative) for our chosen variable against both Social Security Numbers and full names. RESULTS In a data set of 19 000 records, a derived match variable consisting of a 2-character prefix from both first and last names combined with date of birth has a 97% sensitivity; by contrast, an anonymized identifier based on the patients full names and date of birth has a sensitivity of only 87% and SSN has sensitivity 86%. CONCLUSION The approach we describe is most useful in situations where privacy policies preclude the full exchange of the identifiers required by more sophisticated and sensitive linkage algorithms. For data sets of sufficiently high quality this effective approach, while producing a lower rate of matching than more complex algorithms, has the merit of being easy to explain to institutional review boards, adheres to the minimum necessary rule of the HIPAA privacy rule, and is faster and less cumbersome to implement than a full probabilistic linkage.

Multimedia electronic medical record systems.

BACKGROUND The increasing availability of clinical data from electronic medical records (EMRs) has created opportunities for secondary uses of health information. When used in machine learning classification, many data features must first be transformed by discretization. OBJECTIVE To evaluate six discretization strategies, both supervised and unsupervised, using EMR data. MATERIALS AND METHODS We classified laboratory data (arterial blood gas (ABG) measurements) and physiologic data (cardiac output (CO) measurements) derived from adult patients in the intensive care unit using decision trees and naïve Bayes classifiers. Continuous features were partitioned using two supervised, and four unsupervised discretization strategies. The resulting classification accuracy was compared with that obtained with the original, continuous data. RESULTS Supervised methods were more accurate and consistent than unsupervised, but tended to produce larger decision trees. Among the unsupervised methods, equal frequency and k-means performed well overall, while equal width was significantly less accurate. DISCUSSION This is, we believe, the first dedicated evaluation of discretization strategies using EMR data. It is unlikely that any one discretization method applies universally to EMR data. Performance was influenced by the choice of class labels and, in the case of unsupervised methods, the number of intervals. In selecting the number of intervals there is generally a trade-off between greater accuracy and greater consistency. CONCLUSIONS In general, supervised methods yield higher accuracy, but are constrained to a single specific application. Unsupervised methods do not require class labels and can produce discretized data that can be used for multiple purposes.