Stephen T. Wu

Normalization and standardization of electronic health records for high-throughput phenotyping: the SHARPn consortium

RESEARCH OBJECTIVE To develop scalable informatics infrastructure for normalization of both structured and unstructured electronic health record (EHR) data into a unified, concept-based model for high-throughput phenotype extraction. MATERIALS AND METHODS Software tools and applications were developed to extract information from EHRs. Representative and convenience samples of both structured and unstructured data from two EHR systems-Mayo Clinic and Intermountain Healthcare-were used for development and validation. Extracted information was standardized and normalized to meaningful use (MU) conformant terminology and value set standards using Clinical Element Models (CEMs). These resources were used to demonstrate semi-automatic execution of MU clinical-quality measures modeled using the Quality Data Model (QDM) and an open-source rules engine. RESULTS Using CEMs and open-source natural language processing and terminology services engines-namely, Apache clinical Text Analysis and Knowledge Extraction System (cTAKES) and Common Terminology Services (CTS2)-we developed a data-normalization platform that ensures data security, end-to-end connectivity, and reliable data flow within and across institutions. We demonstrated the applicability of this platform by executing a QDM-based MU quality measure that determines the percentage of patients between 18 and 75 years with diabetes whose most recent low-density lipoprotein cholesterol test result during the measurement year was <100 mg/dL on a randomly selected cohort of 273 Mayo Clinic patients. The platform identified 21 and 18 patients for the denominator and numerator of the quality measure, respectively. Validation results indicate that all identified patients meet the QDM-based criteria. CONCLUSIONS End-to-end automated systems for extracting clinical information from diverse EHR systems require extensive use of standardized vocabularies and terminologies, as well as robust information models for storing, discovering, and processing that information. This study demonstrates the application of modular and open-source resources for enabling secondary use of EHR data through normalization into standards-based, comparable, and consistent format for high-throughput phenotyping to identify patient cohorts.

MACE: model based analysis of ChIP-exo

Understanding the role of a given transcription factor (TF) in regulating gene expression requires precise mapping of its binding sites in the genome. Chromatin immunoprecipitation-exo, an emerging technique using λ exonuclease to digest TF unbound DNA after ChIP, is designed to reveal transcription factor binding site (TFBS) boundaries with near-single nucleotide resolution. Although ChIP-exo promises deeper insights into transcription regulation, no dedicated bioinformatics tool exists to leverage its advantages. Most ChIP-seq and ChIP-chip analytic methods are not tailored for ChIP-exo, and thus cannot take full advantage of high-resolution ChIP-exo data. Here we describe a novel analysis framework, termed MACE (model-based analysis of ChIP-exo) dedicated to ChIP-exo data analysis. The MACE workflow consists of four steps: (i) sequencing data normalization and bias correction; (ii) signal consolidation and noise reduction; (iii) single-nucleotide resolution border peak detection using the Chebyshev Inequality and (iv) border matching using the Gale-Shapley stable matching algorithm. When applied to published human CTCF, yeast Reb1 and our own mouse ONECUT1/HNF6 ChIP-exo data, MACE is able to define TFBSs with high sensitivity, specificity and spatial resolution, as evidenced by multiple criteria including motif enrichment, sequence conservation, direct sequence pileup, nucleosome positioning and open chromatin states. In addition, we show that the fundamental advance of MACE is the identification of two boundaries of a TFBS with high resolution, whereas other methods only report a single location of the same event. The two boundaries help elucidate the in vivo binding structure of a given TF, e.g. whether the TF may bind as dimers or in a complex with other co-factors.

A framework for fast incremental interpretation during speech decoding

This article describes a framework for incorporating referential semantic information from a world model or ontology directly into a probabilistic language model of the sort commonly used in speech recognition, where it can be probabilistically weighted together with phonological and syntactic factors as an integral part of the decoding process. Introducing world model referents into the decoding search greatly increases the search space, but by using a single integrated phonological, syntactic, and referential semantic language model, the decoder is able to incrementally prune this search based on probabilities associated with these combined contexts. The result is a single unified referential semantic probability model which brings several kinds of context to bear in speech decoding, and performs accurate recognition in real time on large domains in the absence of example in-domain training sentences.

A common type system for clinical natural language processing

BackgroundOne challenge in reusing clinical data stored in electronic medical records is that these data are heterogenous. Clinical Natural Language Processing (NLP) plays an important role in transforming information in clinical text to a standard representation that is comparable and interoperable. Information may be processed and shared when a type system specifies the allowable data structures. Therefore, we aim to define a common type system for clinical NLP that enables interoperability between structured and unstructured data generated in different clinical settings.ResultsWe describe a common type system for clinical NLP that has an end target of deep semantics based on Clinical Element Models (CEMs), thus interoperating with structured data and accommodating diverse NLP approaches. The type system has been implemented in UIMA (Unstructured Information Management Architecture) and is fully functional in a popular open-source clinical NLP system, cTAKES (clinical Text Analysis and Knowledge Extraction System) versions 2.0 and later.ConclusionsWe have created a type system that targets deep semantics, thereby allowing for NLP systems to encapsulate knowledge from text and share it alongside heterogenous clinical data sources. Rather than surface semantics that are typically the end product of NLP algorithms, CEM-based semantics explicitly build in deep clinical semantics as the point of interoperability with more structured data types.

Coreference analysis in clinical notes: a multi-pass sieve with alternate anaphora resolution modules.

OBJECTIVE This paper describes the coreference resolution system submitted by Mayo Clinic for the 2011 i2b2/VA/Cincinnati shared task Track 1C. The goal of the task was to construct a system that links the markables corresponding to the same entity. MATERIALS AND METHODS The task organizers provided progress notes and discharge summaries that were annotated with the markables of treatment, problem, test, person, and pronoun. We used a multi-pass sieve algorithm that applies deterministic rules in the order of preciseness and simultaneously gathers information about the entities in the documents. Our system, MedCoref, also uses a state-of-the-art machine learning framework as an alternative to the final, rule-based pronoun resolution sieve. RESULTS The best system that uses a multi-pass sieve has an overall score of 0.836 (average of B(3), MUC, Blanc, and CEAF F score) for the training set and 0.843 for the test set. DISCUSSION A supervised machine learning system that typically uses a single function to find coreferents cannot accommodate irregularities encountered in data especially given the insufficient number of examples. On the other hand, a completely deterministic system could lead to a decrease in recall (sensitivity) when the rules are not exhaustive. The sieve-based framework allows one to combine reliable machine learning components with rules designed by experts. CONCLUSION Using relatively simple rules, part-of-speech information, and semantic type properties, an effective coreference resolution system could be designed. The source code of the system described is available at https://sourceforge.net/projects/ohnlp/files/MedCoref.

Negation’s Not Solved: Generalizability Versus Optimizability in Clinical Natural Language Processing

A review of published work in clinical natural language processing (NLP) may suggest that the negation detection task has been “solved.” This work proposes that an optimizable solution does not equal a generalizable solution. We introduce a new machine learning-based Polarity Module for detecting negation in clinical text, and extensively compare its performance across domains. Using four manually annotated corpora of clinical text, we show that negation detection performance suffers when there is no in-domain development (for manual methods) or training data (for machine learning-based methods). Various factors (e.g., annotation guidelines, named entity characteristics, the amount of data, and lexical and syntactic context) play a role in making generalizability difficult, but none completely explains the phenomenon. Furthermore, generalizability remains challenging because it is unclear whether to use a single source for accurate data, combine all sources into a single model, or apply domain adaptation methods. The most reliable means to improve negation detection is to manually annotate in-domain training data (or, perhaps, manually modify rules); this is a strategy for optimizing performance, rather than generalizing it. These results suggest a direction for future work in domain-adaptive and task-adaptive methods for clinical NLP.

Comparative Analysis of Online Health Queries Originating From Personal Computers and Smart Devices on a Consumer Health Information Portal

Background The number of people using the Internet and mobile/smart devices for health information seeking is increasing rapidly. Although the user experience for online health information seeking varies with the device used, for example, smart devices (SDs) like smartphones/tablets versus personal computers (PCs) like desktops/laptops, very few studies have investigated how online health information seeking behavior (OHISB) may differ by device. Objective The objective of this study is to examine differences in OHISB between PCs and SDs through a comparative analysis of large-scale health search queries submitted through Web search engines from both types of devices. Methods Using the Web analytics tool, IBM NetInsight OnDemand, and based on the type of devices used (PCs or SDs), we obtained the most frequent health search queries between June 2011 and May 2013 that were submitted on Web search engines and directed users to the Mayo Clinic’s consumer health information website. We performed analyses on “Queries with considering repetition counts (QwR)” and “Queries without considering repetition counts (QwoR)”. The dataset contains (1) 2.74 million and 3.94 million QwoR, respectively for PCs and SDs, and (2) more than 100 million QwR for both PCs and SDs. We analyzed structural properties of the queries (length of the search queries, usage of query operators and special characters in health queries), types of search queries (keyword-based, wh-questions, yes/no questions), categorization of the queries based on health categories and information mentioned in the queries (gender, age-groups, temporal references), misspellings in the health queries, and the linguistic structure of the health queries. Results Query strings used for health information searching via PCs and SDs differ by almost 50%. The most searched health categories are “Symptoms” (1 in 3 search queries), “Causes”, and “Treatments & Drugs”. The distribution of search queries for different health categories differs with the device used for the search. Health queries tend to be longer and more specific than general search queries. Health queries from SDs are longer and have slightly fewer spelling mistakes than those from PCs. Users specify words related to women and children more often than that of men and any other age group. Most of the health queries are formulated using keywords; the second-most common are wh- and yes/no questions. Users ask more health questions using SDs than PCs. Almost all health queries have at least one noun and health queries from SDs are more descriptive than those from PCs. Conclusions This study is a large-scale comparative analysis of health search queries to understand the effects of device type (PCs vs SDs) used on OHISB. The study indicates that the device used for online health information search plays an important role in shaping how health information searches by consumers and patients are executed.

A Hybrid Approach to Sentiment Sentence Classification in Suicide Notes

This paper describes the sentiment classification system developed by the Mayo Clinic team for the 2011 I2B2/VA/Cincinnati Natural Language Processing (NLP) Challenge. The sentiment classification task is to assign any pertinent emotion to each sentence in suicide notes. We have implemented three systems that have been trained on suicide notes provided by the I2B2 challenge organizer–-a machine learning system, a rule-based system, and a system consisting of a combination of both. Our machine learning system was trained on re-annotated data in which apparently inconsistent emotion assignment was adjusted. Then, the machine learning methods by RIPPER and multinomial Naïve Bayes classifiers, manual pattern matching rules, and the combination of the two systems were tested to determine the emotions within sentences. The combination of the machine learning and rule-based system performed best and produced a micro-average F-score of 0.5640.

Patient-level temporal aggregation for text-based asthma status ascertainment

OBJECTIVE To specify the problem of patient-level temporal aggregation from clinical text and introduce several probabilistic methods for addressing that problem. The patient-level perspective differs from the prevailing natural language processing (NLP) practice of evaluating at the term, event, sentence, document, or visit level. METHODS We utilized an existing pediatric asthma cohort with manual annotations. After generating a basic feature set via standard clinical NLP methods, we introduce six methods of aggregating time-distributed features from the document level to the patient level. These aggregation methods are used to classify patients according to their asthma status in two hypothetical settings: retrospective epidemiology and clinical decision support. RESULTS In both settings, solid patient classification performance was obtained with machine learning algorithms on a number of evidence aggregation methods, with Sum aggregation obtaining the highest F1 score of 85.71% on the retrospective epidemiological setting, and a probability density function-based method obtaining the highest F1 score of 74.63% on the clinical decision support setting. Multiple techniques also estimated the diagnosis date (index date) of asthma with promising accuracy. DISCUSSION The clinical decision support setting is a more difficult problem. We rule out some aggregation methods rather than determining the best overall aggregation method, since our preliminary data set represented a practical setting in which manually annotated data were limited. CONCLUSION Results contrasted the strengths of several aggregation algorithms in different settings. Multiple approaches exhibited good patient classification performance, and also predicted the timing of estimates with reasonable accuracy.

Using Empirically Constructed Lexical Resources for Named Entity Recognition

Because of privacy concerns and the expense involved in creating an annotated corpus, the existing small-annotated corpora might not have sufficient examples for learning to statistically extract all the named-entities precisely. In this work, we evaluate what value may lie in automatically generated features based on distributional semantics when using machine-learning named entity recognition (NER). The features we generated and experimented with include n-nearest words, support vector machine (SVM)-regions, and term clustering, all of which are considered distributional semantic features. The addition of the n-nearest words feature resulted in a greater increase in F-score than by using a manually constructed lexicon to a baseline system. Although the need for relatively small-annotated corpora for retraining is not obviated, lexicons empirically derived from unannotated text can not only supplement manually created lexicons, but also replace them. This phenomenon is observed in extracting concepts from both biomedical literature and clinical notes.