Faculty Bibliography

The clinical research infrastructure of the United States is currently at a critical crossroads. To leverage the enormous biomedical research gains made in the past century efficiently, a drastic need exists to reengineer this system into a coordinated, safe, and more efficient and effective enterprise. To accomplish this task, clinical research must be transformed from its current state as a cottage industry to an enterprise-wide health care pipeline whose function is to bring the novel research from both government and private entities to the US public. We propose the establishment of a unique public-private partnership termed the National Clinical Research Enterprise (NCRE). Its agenda should consist of informed public participation, supportive information technologies, a skilled workforce, and adequate funding in clinical research. Devoting only 0.25% of the budgets from all health care stakeholders to support the NCRE would permit adequate funding to build the infrastructure required to address these problems in an enterprise fashion. All participants in the US health care delivery system must come together to focus on system-wide improvements that will benefit the public.

Physicians' information needs routinely arise during their practice. Several studies have demonstrated that a useful way to understand the nature of the needs is to examine questions posed by physicians during the course of medical care. This paper presents an analysis of clinical questions collected when physicians were engaged in reviewing clinical cases. The main objective of the study was to investigate characteristics of the physicians' information needs to provide insight into the development of a conceptual guidance approach in information retrieval. The analysis focused on categories, search contexts, and patterns of the questions with respect to the scenarios. The results of the study showed that physicians' information needs exhibited distinct characteristics according to the scenarios, and that their needs could be expressed with a relatively small number of question patterns. These observations supported our approach of using patterns of information needs in facilitating digital information access.

Many interventions to improve the success of information technology (IT) implementations are grounded in behavioral science, using theories, and models to identify conditions and determinants of successful use. However, each model in the IT literature has evolved to address specific theoretical problems of particular disciplinary concerns, and each model has been tested and has evolved using, in most cases, a more or less restricted set of IT implementation procedures. Functionally, this limits the perspective for taking into account the multiple factors at the individual, group, and organizational levels that influence use behavior. While a rich body of literature has emerged, employing prominent models such as the Technology Adoption Model, Social-Cognitive Theory, and Diffusion of Innovation Theory, the complexity of defining a suitable multi-level intervention has largely been overlooked. A gap exists between the implementation of IT and the integration of theories and models that can be utilized to develop multi-level approaches to identify factors that impede usage behavior. We present a novel framework that is intended to guide synthesis of more than one theoretical perspective for the purpose of planning multi-level interventions to enhance IT use. This integrative framework is adapted from PRECEDE/PROCEDE, a conceptual framework used by health planners in hundreds of published studies to direct interventions that account for the multiple determinants of behavior. Since we claim that the literature on IT use behavior does not now include a multi-level approach, we undertook a systematic literature analysis to confirm this assertion. Our framework facilitated organizing this literature synthesis and our analysis was aimed at determining if the IT implementation approaches in the published literature were characterized by an approach that considered at least two levels of IT usage determinants. We found that while 61% of studies mentioned or referred to theory, none considered two or more levels. In other words, although the researchers employ behavioral theory, they omit two fundamental propositions: (1) IT usage is influenced by multiple factors and (2) interventions must be multi-dimensional. Our literature synthesis may provide additional insight into the reason for high failure rates associated with underutilized systems, and underscores the need to move beyond the current dominant approach that employs a single model to guide IT implementation plans that aim to address factors associated with IT acceptance and subsequent positive use behavior.

Medical scientists and public health policy makers are increasingly concerned that the scientific discoveries of the past generation are failing to be translated efficiently into tangible human benefit. This concern has generated several initiatives, including the Clinical Research Roundtable at the Institute of Medicine, which first convened in June 2000. Representatives from a diverse group of stakeholders in the nation's clinical research enterprise have collaborated to address the issues it faces. The context of clinical research is increasingly encumbered by high costs, slow results, lack of funding, regulatory burdens, fragmented infrastructure, incompatible databases, and a shortage of qualified investigators and willing participants. These factors have contributed to 2 major obstacles, or translational blocks: impeding the translation of basic science discoveries into clinical studies and of clinical studies into medical practice and health decision making in systems of care. Considering data from across the entire health care system, it has become clear that these 2 translational blocks can be removed only by the collaborative efforts of multiple system stakeholders. The goal of this article is to articulate the 4 central challenges facing clinical research at present--public participation, information systems, workforce training, and funding; to make recommendations about how they might be addressed by particular stakeholders; and to invite a broader, participatory dialogue with a view to improving the overall performance of the US clinical research enterprise.

The problem of developing a curriculum for biomedical informatics is highly dependent on how we choose to define and practice the field. Numerous authors have questioned how to position biomedical informatics along the continuum of formal, empirical and engineering disciplines. A concern with current educational programs in biomedical informatics is that students finish without a clear understanding of the relation between theory and practice, or worse, with the impression that the field does not possess any theoretical basis. In this paper, we propose that biomedical informatics curricula explicitly address skills and competencies at three levels: formal, empirical, and applied. We posit that that knowledge of formalization is necessary to build testable empirical models, and that model-driven approaches are necessary for deploying information systems that can be evaluated in a meaningful way. A curricular framework is proposed that identifies a set of methods, techniques and theories that have broad applicability within the domain of biomedicine, and which can span a wide range of application areas: bioinformatics, imaging informatics, clinical informatics and public health informatics. A stronger linkage between theory and practice will result in students who are empowered to create effective and lasting solutions to biomedical problems.

In recent years, significant progress has been achieved toward increased structured data entry using standardized health care terminologies. Concurrently, the value of narrative as the clinician's rich description of the encounter and source of vital information has been reaffirmed. Natural language processing (NLP) offers a strategy for integrating these approaches to provide structured reports for further computer processing. As part of a larger project aimed at using narrative data to enrich the online medical record, we analyzed a small sample of documents in a corpus of progress notes to identify potential challenges associated with using NLP for HIV/AIDS clinic notes. We provide illustrative examples of five types of challenges.

Health-care institutions are gaining an increasing interest in exploiting the data that are gathered through electronic medical records. Narrative data, generated by transcription or direct entry, represents a far greater challenge for analytic tasks. Moreover, a small number of institutions are beginning to explore deeper structuring of narrative data using natural language processing (NLP). The data produced by NLP systems has a complex, nested structure. Current electronic medical records do not have the ability to store and retrieve data of this complexity in a suitable way.

Conversion of free-text strings in a natural language to a standard representation (codes) is an important reoccurring problem in biomedical informatics. Determining the content of a string involves identifying its meaningful constituents (morphemes). One current method of identifying these constituents is to look them up in a preexisting table (lexicon). Manual construction of lexicons and grammars in complex domains such as biomedicine is extremely laborious. As an alternative to the lexico-grammatical approach, we introduce a segmentation algorithm that automatically learns lexical and structural preferences from corpora via information compression. The method is based on the Minimum Description Length (MDL) principle from classic information theory.

UNLABELLED:Motivation. The UMLS has been used in natural language processing applications such as information retrieval and information extraction systems. The mapping of free-text to UMLS concepts is important for these applications. To improve the mapping, we need a method to disambiguate terms that possess multiple UMLS concepts. In the general English domain, machine-learning techniques have been applied to sense-tagged corpora, in which senses (or concepts) of ambiguous terms have been annotated (mostly manually). Sense disambiguation classifiers are then derived to determine senses (or concepts) of those ambiguous terms automatically. However, manual annotation of a corpus is an expensive task. We propose an automatic method that constructs sense-tagged corpora for ambiguous terms in the UMLS using MEDLINE abstracts. METHODS:For a term W that represents multiple UMLS concepts, a collection of MEDLINE abstracts that contain W is extracted. For each abstract in the collection, occurrences of concepts that have relations with W as defined in the UMLS are automatically identified. A sense-tagged corpus, in which senses of W are annotated, is then derived based on those identified concepts. The method was evaluated on a set of 35 frequently occurring ambiguous biomedical abbreviations using a gold standard set that was automatically derived. The quality of the derived sense-tagged corpus was measured using precision and recall. RESULTS:The derived sense-tagged corpus had an overall precision of 92.9% and an overall recall of 47.4%. After removing rare senses and ignoring abbreviations with closely related senses, the overall precision was 96.8% and the overall recall was 50.6%. CONCLUSIONS:UMLS conceptual relations and MEDLINE abstracts can be used to automatically acquire knowledge needed for resolving ambiguity when mapping free-text to UMLS concepts.

XML has been widely adopted as an important data interchange language. The structure of XML enables sharing of data elements with variable degrees of nesting as long as the elements are grouped in a strict tree-like fashion. This requirement potentially restricts the usefulness of XML for marking up written text, which often includes features that do not properly nest within other features. We encountered this problem while marking up medical text with structured semantic information from a Natural Language Processor. Traditional approaches to this problem separate the structured information from the actual text mark up. This paper introduces an alternative solution, which tightly integrates the semantic structure with the text. The resulting XML markup preserves the linearity of the medical texts and can therefore be easily expanded with additional types of information.