Patient safety and emergency department functionality are compromised when inefficient coordination between hospital departments impedes ED patients access to inpatient cardiac care. The objective of this dissertation was to create a discrete event simulation model of hospital patient flow in order to determine how bed demand from competing cardiology admission sources affects ED patients access to inpatient cardiac care. The simulation employed survival analysis regression to model competition for inpatient cardiac beds and predict delays in ED patients access. The novel simulation strategy was used to demonstrate how altering outpatient schedules and creating informed bed management practices can optimize hospital capacity and improve ED patient access and how interventions designed to increase inpatient throughput or add capacity will have the most significant effect on highest priority patients.

Previous studies have reported that fluorescence in human tissue is a temperature dependent phenomenon. The most apparent effect is the inverse relationship between fluorescence intensity and temperature. Here, we present the effects of temperature on fluorescence in human tissue using skin and adipose tissue in vitro and skin in vivo. Fluorescence and diffuse reflectance measurements were made while the temperature of the specimens was increased, and an inverse Monte Carlo algorithm was used to calculate optical properties as a function of temperature. In vitro and in vivo experiments showed a decrease in fluorescence intensity due to temperature increase. The fluorescence intensity showed no relationship to the optical properties in the physiological temperature range, suggesting that changes in optical properties are not the primary mechanism by which fluorescence is affected by temperature. This study confirms that fluorescence decreases with increasing temperature in human tissue in vitro and in vivo. Our results further suggest the presence of a temperature dependent non-radiative decay mechanism.

<p>Answering questions such as Which genes are related to breast cancer? usually requires retrieving relevant publications through the PubMed search engine, reading these publications, and manually creating gene lists. This process is both time-consuming and prone to errors. </p> <p>We report GLAD4U (Gene List Automatically Derived For You), a novel, free web-based gene retrieval and prioritization tool. The quality of gene lists created by GLAD4U for three Gene Ontology terms and three disease terms was assessed using gold standard lists curated in public databases. We also compared the performance of GLAD4U with that of another gene prioritization software, EBIMed. </p> <p>GLAD4U has a high overall recall. Although precision is generally low, its prioritization methods successfully rank truly relevant genes at the top of generated lists to facilitate efficient browsing. GLAD4U is simple to use, and its interface can be found at: http://bioinfo.vanderbilt.edu/glad4u.</p>

By sampling the self-diffusion of water molecules, diffusion tensor imaging (DTI) is able to characterize the microstructure of brain white matter. Previous DTI studies in schizophrenia have reported white matter alterations as measured by changes in fractional anisotropy. However, DTI analysis is not capable of distinguishing between possible causes, such as a change in the fiber orientation coherence, a change in the intrinsic diffusivity of the fibers, or both. Compared with DTI, high angular resolution diffusion imaging (HARDI) provides more detailed structural information of underlying tissues. Fiber ORientation Estimated using Continuous Axially Symmetric Tensors (FORECAST) is a spherical deconvolution method to analyze HARDI data.<p> Based on Monte Carlo simulations, as well as bootstrap analysis of in vivo human data, the optimal imaging and processing parameters for conducting the FORECAST analysis within typical clinical constraints were determined, and the accuracy of the model was estimated. <p> In order to compare HARDI measurements between subjects, an algorithm was developed to transform the fiber orientation distribution (FOD) function, based on HARDI data, taking into account not only translation, but also rotation, scaling, and shearing effects of the spatial transformation. The algorithm was tested using simulated data, and intra-subject and inter-subject normalization of in vivo human data. All cases demonstrate reliable transformation of the FOD. <p> A voxel-based group comparison of the radial diffusivity and intravoxel fiber coherence was performed based on FORECAST analysis of the HARDI images from both healthy controls and patients with schizophrenia. Decreased FA and elevated radial diffusivity were found in a number of white matter regions in patients. Our results suggest that increased radial diffusivity is the major contributor to the FA reduction, while decreased intravoxel fiber coherence also plays a role in the white matter alterations. This set of techniques, as a step forward from conventional DTI analysis, will likely be helpful in clinical studies of other white matter diseases as well.

This Masters Thesis project had as its objectives: (1) to provide Vanderbilt University Hospital (VUH) with computerized tools for monitoring microbiological data; (2) to provide the VUH Infection Control Service with tools to help monitor and track infection-relevant patient-related data such as culture results, hospital location, current orders, and contact precautions status; and, (3) to initiate studies to improve compliance with VUH contact precautions policies specifically, those for antibiotic-resistant organisms such as methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus (VRE). This project achieved its goals by developing and formatively evaluating the MicroTools suite of programs. MicroParse processes VUH microbiology laboratory reports, MicroDash provides infection control staff with aggregated information on patients with a history of antibiotic-resistant infections, and MicroGram generates antibiograms for VUH clinicians.

The aims of this project are to 1) build and evaluate a terminology that provides categorization labels, or tags, for common segments within clinical documents, and 2) to evaluate a tool to parse and label natural-language clinical documents using the terminology. Clinical documents generally contain many sections and subsections, such as history of present illness, physical examination, and cardiovascular exam. The author developed a section header terminology that models common section names, subsection names, and their relationships. This terminology was built using existing standardized terminologies, textbooks, and review of over 9,000 clinical notes. The section tagging tool, named SecTag, identifies terminology matches from clinical documents using a combination of linguistic, natural language processing, and machine learning techniques. The evaluation study focused on recognizing sections in 319 randomly-chosen history and physical examination notes that were generated during hospitalizations and outpatient visits. The overall recall and precision were 99% and 96%, respectively, over 16,036 possible sections. Recall and precision for sections not labeled in the document were 97% and 87%, respectively. The system correctly tagged 93% of the section start and end boundaries. SecTag failed to label 160 sections (1%); only 11 were headings that were absent in the terminology and which should be added to it. SecTag and its terminology are important first steps for understanding clinical notes. Future studies are needed to extend the terminology to other clinical note types and to link SecTag to a more in-depth natural language processing system.

The presence of tumor within 1-2 millimeters of the surgical margin following partial mastectomies is strongly correlated with the risk of local breast tumor recurrence; thus, there is a need to develop a non-invasive, real-time, accurate margin evaluation tool to assure complete tumor removal. We previously showed that both Raman spectroscopy and combined autofluorescence and diffuse reflectance spectroscopy can accurately discriminate normal from malignant breast tissues. In this work, the successful use of fluorescence and reflectance-based spectroscopy and spectral imaging for superficial margin evaluation was demonstrated. To achieve greater depth sampling, spatially offset Raman spectroscopy (SORS) was implemented. The feasibility of SORS for breast surgical margin analysis was established by detecting breast tumor signatures through a maximum of 2 mm of normal human breast tissue. A SORS Monte Carlo simulation model was then developed to investigate the effects of varying both normal and tumor layer thicknesses over a wide range of values. The experimental and theoretical SORS results were then used to design a multi-separation SORS probe capable of detecting tumor signatures from below a maximum of 2 mm of normal breast tissue. This probe was used to acquire Raman spectra from frozen-thawed normal breast and breast tumor samples in the laboratory, and a probabilistic classification scheme was developed to determine whether any tumor signature was present in the first 2 mm of tissue under the probe site. Measurements were then made on a small set of freshly excised breast specimens in the clinic to ensure the feasibility of translating this technique to the operating room.

In this dissertation, I explore the applicability of text categorization machine learning methods to identify clinically pertinent and evidence-based articles in the literature and web pages on the internet. In the first series of experiments, I found that text categorization techniques identify high quality articles in internal medicine in the content categories of prognosis, diagnosis, etiology, and treatment better than the Clinical Query Filters of Pubmed. In a second set of experiments, I established that the text categorization models generalized both to time periods outside the training set and to areas outside of internal medicine including pediatrics, oncology, and surgery. My third set of experiments revealed that text categorization models built for a specific purpose identified articles better than both bibliometric (number of citations and impact factor) and web-based measures (Google PageRank, Yahoo WebRanks, and total web page hit count). In the fourth set of experiments, I built models for purpose, format, and additional content categories from a labeled gold standard that have high discriminatory power. Furthermore, we built a system called EBMSearch that implements these models to all of MEDLINE. Finally I extended these methods to the web and built the first validated models that identify websites that make false cancer treatment claims outperforming previous unvalidated models and PageRank by 30% area under the receiver operating curve. In conclusion, machine learning-based text categorization methods provide a powerful framework for identifying clinically applicable articles in the medical literature and the Internet.

Clinical decision support systems can decrease common errors related to inadequate dosing for nephrotoxic or renally cleared drugs. Within the computerized provider order entry (CPOE) system, we developed, implemented, and evaluated a set of interventions with varying levels of workflow intrusiveness to continuously monitor for and alert providers about acute kidney injury. Passive alerts appeared as persistent text within the CPOE system and on rounding reports, requiring no provider response. Exit check alerts interrupted the provider at the end of the CPOE session, requiring the provider to modify or discontinue the drug order, assert the current dose as correct, or defer the alert. In the intervention period, the number of drugs modified or discontinued within 24 hours increased from 35.7% to 50.9%, and the median time to modification or discontinuation decreased from 27.1 hours to 12.9 hours. Providers delayed decisions by repeatedly deferring the alerts. Future enhancements will address frequent deferrals by involving other team members in making mid-regimen prescription decisions.

Neural stimulation using infrared light has recently been characterized as a novel method to stimulate peripheral nerves without touching, causing damage, or inducing an electrical stimulation artifact. Infrared neural stimulation (INS) has not been previously achieved in the brain due to the complexity of the neuronal networks. The purpose of this study was to show feasibility of INS in the central nervous system, and determine the optimal parameters for INS in a thalamocortical brain slice model. Wavelength was the first parameter identified since previous studies showed penetration depth of light in tissue determined the threshold radiant energies needed to evoke an action potential in the peripheral nervous system. The wavelength of 3.65 &181m was determined to be the optimal wavelength. Next repetition rate was investigated using the optimal wavelength of 3.65 &181m. Lower threshold radiant energies were observed for higher repetition rates. The final parameter investigated was spot size using light at 3.65 &181m and 30 Hz, and a third order power fit relationship was observed where a larger spot size required less energy to evoke action potentials in a TC slice. A small set of experiments were performed to show intracellular electrical recordings could be used to detect INS evoked signals. The results from this study prove feasibility of INS in CNS, and provide the basis for future in vivo experiments.