BIOMEDICAL LANGUAGE UNDERSTANDING AND EXTRACTION (BLUE-TEXT): A MINIMAL SYNTACTIC, SEMANTIC METHOD

Clinical text understanding (CTU) is of interest to health informatics because critical clinical information frequently represented as unconstrained text in electronic health records are extensively used by human experts to guide clinical practice, decision making, and to document delivery of care, but are largely unusable by information systems for queries and computations. Recent initiatives advocating for translational research call for generation of technologies that can integrate structured clinical data with unstructured data, provide a unified interface to all data, and contextualize clinical information for reuse in multidisciplinary and collaborative environment envisioned by CTSA program. This implies that technologies for the processing and interpretation of clinical text should be evaluated not only in terms of their validity and reliability in their intended environment, but also in light of their interoperability, and ability to support information integration and contextualization in a distributed and dynamic environment. This vision adds a new layer of information representation requirements that needs to be accounted for when conceptualizing implementation or acquisition of clinical text processing tools and technologies for multidisciplinary research. On the other hand, electronic health records frequently contain unconstrained clinical text with high variability in use of terms and documentation practices, and without commitmentto grammatical or syntactic structure of the language (e.g. Triage notes, physician and nurse notes, chief complaints, etc). This hinders performance of natural language processing technologies which typically rely heavily on the syntax of language and grammatical structure of the text. This document introduces our method to transform unconstrained clinical text found in electronic health information systems to a formal (computationally understandable) representation that is suitable for querying, integration, contextualization and reuse, and is resilient to the grammatical and syntactic irregularities of the clinical text. We present our design rationale, method, and results of evaluation in processing chief complaints and triage notes from 8 different emergency departments in Houston Texas. At the end, we will discuss significance of our contribution in enabling use of clinical text in a practical bio-surveillance setting.

Data Accuracy in Medical Record Abstraction

Clinical Research Data Quality Literature Review and Pooled Analysis We present a literature review and secondary analysis of data accuracy in clinical research and related secondary data uses. A total of 93 papers meeting our inclusion criteria were categorized according to the data processing methods. Quantitative data accuracy information was abstracted from the articles and pooled. Our analysis demonstrates that the accuracy associated with data processing methods varies widely, with error rates ranging from 2 errors per 10,000 files to 5019 errors per 10,000 fields. Medical record abstraction was associated with the highest error rates (70–5019 errors per 10,000 fields). Data entered and processed at healthcare facilities had comparable error rates to data processed at central data processing centers. Error rates for data processed with single entry in the presence of on-screen checks were comparable to double entered data. While data processing and cleaning methods may explain a significant amount of the variability in data accuracy, additional factors not resolvable here likely exist. Defining Data Quality for Clinical Research: A Concept Analysis Despite notable previous attempts by experts to define data quality, the concept remains ambiguous and subject to the vagaries of natural language. This current lack of clarity continues to hamper research related to data quality issues. We present a formal concept analysis of data quality, which builds on and synthesizes previously published work. We further posit that discipline-level specificity may be required to achieve the desired definitional clarity. To this end, we combine work from the clinical research domain with findings from the general data quality literature to produce a discipline-specific definition and operationalization for data quality in clinical research. While the results are helpful to clinical research, the methodology of concept analysis may be useful in other fields to clarify data quality attributes and to achieve operational definitions. Medical Record Abstractor’s Perceptions of Factors Impacting the Accuracy of Abstracted Data Medical record abstraction (MRA) is known to be a significant source of data errors in secondary data uses. Factors impacting the accuracy of abstracted data are not reported consistently in the literature. Two Delphi processes were conducted with experienced medical record abstractors to assess abstractor’s perceptions about the factors. The Delphi process identified 9 factors that were not found in the literature, and differed with the literature by 5 factors in the top 25%. The Delphi results refuted seven factors reported in the literature as impacting the quality of abstracted data. The results provide insight into and indicate content validity of a significant number of the factors reported in the literature. Further, the results indicate general consistency between the perceptions of clinical research medical record abstractors and registry and quality improvement abstractors. Distributed Cognition Artifacts on Clinical Research Data Collection Forms Medical record abstraction, a primary mode of data collection in secondary data use, is associated with high error rates. Distributed cognition in medical record abstraction has not been studied as a possible explanation for abstraction errors. We employed the theory of distributed representation and representational analysis to systematically evaluate cognitive demands in medical record abstraction and the extent of external cognitive support employed in a sample of clinical research data collection forms. We show that the cognitive load required for abstraction in 61% of the sampled data elements was high, exceedingly so in 9%. Further, the data collection forms did not support external cognition for the most complex data elements. High working memory demands are a possible explanation for the association of data errors with data elements requiring abstractor interpretation, comparison, mapping or calculation. The representational analysis used here can be used to identify data elements with high cognitive demands.

Verb generation in children with spina bifida.

We investigated verb generation in children with spina bifida meningomyelocele (SBM; n = 55) and in typically developing controls (n = 32). Participants completed 6 blocks (40 trials each) of a task requiring them to produce a semantically related verb in response to a target noun and an additional 40 trials on which they were simply required to read target nouns aloud. After controlling for reading response time, groups did not differ significantly in verb generation response time or learning. Children with SBM produced more non-verb errors than controls and tended to repeat their mistakes over blocks. Verb generation performance was associated with brain volume measures in participants with SBM. Congenital cerebellar dysmorphology is associated with impaired performance in verb generation accuracy, although not with increased response times to produce verbs

Defining the role of IL-15 trans-presentation by distinct cell-types during the development and homeostasis of Natural Killer and invariant Natural Killer T cells

The immuno-regulatory functions displayed by NK and iNKT cells have highlighted their importance as key lymphocytes involved in innate and adaptive immunity. Therefore, understanding the dynamics influencing the generation of NK and iNKT cells is extremely important. IL-15 has been shown to provide a critical signal throughout the development and homeostasis of NK and iNKT cells; however, the cellular source of IL-15 has remained unclear. In this investigation, I provide evidence that the cell-type providing IL-15 to NK and iNKT cells via trans-presentation is determined by the tissue site and the maturation status of NK and iNKT cells. For NK cells, I revealed the non-hematopoietic compartment provides IL-15 to NK cells in the early stages of development while hematopoietic cells were crucial for the generation and maintenance of mature NK cells. Regarding iNKT cells in the thymus, IL-15 trans-presentation by non-hematopoietic cells was crucial for the survival of mature iNKT cells. In the liver, both hematopoietic and non-hematopoietic compartments provided IL-15 to both immature and mature iNKT cells. This IL-15 signal helped mediate the survival and proliferation of both NK and iNKT cells as well as induce the functional maturation of mature iNKT cells via enhanced T-bet expression. In conclusion, my work illustrates an important notion that the immunological niche of NK and iNKT cells is tightly regulated and that this regulation is meticulously influenced by the tissue microenvironment.

Mechanism of surgical stress impairment of murine natural killer cell cytotoxicity

Natural killer cells may provide an important first line of defense against metastatic implantation of solid tumors. This antitumor function occurs during the intravascular and visceral lodgment phase of cancer dissemination, as demonstrated in small animal metastasis models. The role of the NK cell in controlling human tumor dissemination is more difficult to confirm, at least partially because of ethical restraints on experimental design. Nonetheless, a large number of solid tumor patient studies have demonstrated NK cell cytolysis of both autologous and allogeneic tumors.^ Of the major cancer therapeutic modalities, successful surgery in conjunction with other treatments offers the best possibility of cure. However, small animal experiments have demonstrated that surgical stress can lead to increased rates of primary tumor take, and increased incidence, size, and rapidity of metastasis development. Because the physiologic impact of surgical stress can also markedly impair perioperative antitumor immune function in humans, we examined the effect of surgical stress on perioperative NK cell cytolytic function in a murine preclinical model. Our studies demonstrated that hindlimb amputation led to a marked impairment of postoperative NK cell cytotoxicity. The mechanism underlying this process is complex and involves the postsurgical generation of splenic erythroblasts that successfully compete with NK cells for tumor target binding sites; NK cell-directed suppressor cell populations; and a direct impairment of NK cell recycling capacity. The observed postoperative NK cell suppression could be prevented by in vivo administration of pyrimidinone biologic response modifiers or by short term in vitro exposure of effector cells to recombinant Interleukin-2. It is hoped that insights gained from this research may help in the future development of NK cell specific perioperative immunotherapy relevant to the solid tumor patients undergoing cancer resection. ^

ANALYSIS OF THE CORYNEBACTERIUM PARVUM-INDUCED DECLINE OF MURINE NATURAL KILLER CELL-MEDIATED CYTOTOXICITY

Treatment of mice with the immunomodulating agent, Corynebacterium parvum (C. parvum), was shown to result in a severe and long-lasting depression of splenic natural killer (NK) cell-mediated cytotoxicity 5-21 days post-inoculation. Because NK cells have been implicated in immunosurveillance against malignancy (due to their spontaneous occurrence and rapid reactivity to a variety of histological types of tumors), as well as in resistance to established tumors, this decreased activity was of particular concern, since this effect is contrary to that which would be considered therapeutically desirable in cancer treatment (i.e. a potentiation of antitumor effector functions, including NK cell activity, would be expected to lead to a more effective destruction of malignant cells). Therefore, an analysis of the mechanism of this decline of splenic NK cell activity in C.parvum treated mice was undertaken.^ From in vitro co-culturing experiments, it was found that low NK-responsive C. parvum splenocytes were capable of reducing the normally high-reactivity of cells from untreated syngeneic mice to YAC-1 lymphoma, suggesting the presence of NK-directed suppressor cells in C. parvum treated animals. This was further supported by the demonstration of normal levels of cytotoxicity in C. parvum splenocyte preparations following Ficoll-Hypaque separation, which coincided with removal of the NK-suppressive capabilities of these cells. The T cell nature of these regulatory cells was indicated by (1) the failure of C. parvum to cause a reduction of NK cell activity, or the generation of NK-directed suppressor cells in T cell-deficient athymic mice, (2) the removal of C. parvum-induced suppression by T cell-depleting fractionation procedures or treatments, and (3) demonstration of suppression of NK cell activity by T cell-enriched C. parvum splenocytes. These studies suggest, therefore, that the eventual reduction of suppression by T cell elimination and/or inhibition, may result in a promotion of the antitumor effectiveness of C. parvum due to the contribution of "freed" NK effector cell activity.^ However, the temporary suppression of NK cell activity induced by C. parvum (reactivity of treated mice returns to normal levels within 28 days after C. parvum injection), may in fact be favorable in some situations, e.g. in bone marrow transplantation cases, since NK cells have been suggested to play a role also in the process of bone marrow graft rejection.^ Therefore, the discriminate use of agents such as C. parvum may allow for the controlled regulation of NK cell activity suggested to be necessary for the optimalization of therapeutic regimens. ^

Network-based analysis of genome-wide scans of natural selection

The genomic era brought by recent advances in the next-generation sequencing technology makes the genome-wide scans of natural selection a reality. Currently, almost all the statistical tests and analytical methods for identifying genes under selection was performed on the individual gene basis. Although these methods have the power of identifying gene subject to strong selection, they have limited power in discovering genes targeted by moderate or weak selection forces, which are crucial for understanding the molecular mechanisms of complex phenotypes and diseases. Recent availability and rapid completeness of many gene network and protein-protein interaction databases accompanying the genomic era open the avenues of exploring the possibility of enhancing the power of discovering genes under natural selection. The aim of the thesis is to explore and develop normal mixture model based methods for leveraging gene network information to enhance the power of natural selection target gene discovery. The results show that the developed statistical method, which combines the posterior log odds of the standard normal mixture model and the Guilt-By-Association score of the gene network in a naïve Bayes framework, has the power to discover moderate/weak selection gene which bridges the genes under strong selection and it helps our understanding the biology under complex diseases and related natural selection phenotypes.^