Polymer properties can be altered by copolymerizing subunits with specific physicochemical characteristics. Vascular stent materials require biocompatibility, mechanical strength, and prevention of restenosis. Here we copolymerized poly(ε-caprolactone) (PCL), poly(ethylene glycol) (PEG), and carboxyl-PCL (cPCL) at varying molar ratios and characterized the resulting effects on physicochemical and mechanical properties. We then evaluated these polymers for their applicability as coronary stent materials using two primary human coronary artery cell types: smooth muscle cells (HCASMCs) and endothelial cells (HCAECs). Changes of their proliferation and phenotype were dependent upon intracellular reactive oxygen species (ROS) levels, and 4%PEG-96%PCL was identified as the most appropriate material for this application. On this substrate, HCASMCs maintained a contractile phenotype identified by arrested proliferation, moderate oxidative activity, up-regulated expression of smooth muscle myosin heavy chain (smMHC), and healthy spindle morphology. HCAECs on 4%PEG-96%PCL maintained a physiologically-relevant proliferation rate and a balanced redox state. Other test substrates promoted a pathological, synthetic phenotype in HCASMCs and/or hyperproliferation in HCAECs. The cellular responses related to the phenotypic change were modulated by Youngs modulus and surface charge of test substrates, indicating a structure-function relationship that can be exploited for intricate control over vascular cell functions.

The recent development of low-intensity, pulsed infrared light for neural activation has provided a new nerve stimulation modality that avoids the limitations of traditional electrical methods such as the necessity of contact, presence of a stimulation artifact and poor spatial precision. Infrared neural stimulation is, however, limited by a 2:1 ratio of damaging radiant exposures to stimulation threshold radiant exposures. For infrared neural stimulation to become more applicable and eventually suitable for implantation, the range of safe and effective radiant exposures as indicated by this ratio must be increased. In this study, we have shown that this ratio is increased to as much as 7:1 by combining the infrared pulse with a subthreshold depolarizing electrical stimulus. Our results indicate a nonlinear relationship between the subthreshold depolarizing electrical stimulus (expressed as percentage of electrical stimulation threshold) and the additional optical energy required to reach stimulation threshold (expressed as percentage of optical stimulation threshold). The results also show that the change in optical threshold decreases linearly as the delay between the electrical and optical pulses is increased. The primary benefit of infrared neural stimulation is spatial selectivity and we have shown that precision is maintained for this combined stimulation modality. Our findings are evaluated in the context of latent addition and "superexcitability" according to previously published results. The results of this study are expected to facilitate the development of applications for infrared neural stimulation, as well as target the efforts to uncover the mechanism by which infrared light activates neural tissue.

Many medications have potentially serious adverse effects detected only after FDA approval. After 80 million people worldwide received prescriptions for the drug rofecoxib (Vioxx), its manufacturer withdrew it from the marketplace in 2004. Epidemiological data showed that it increases risk of heart attack and stroke. Recently, the FDA warned that the commonly prescribed statin drug class (e.g., Lipitor, Zocor, Crestor) may increase risk of memory loss and Type 2 diabetes. These incidents illustrate the difficulty of identifying adverse effects of prescription medications during premarketing trials. Only post-marketing surveillance can detect some types of adverse effects (e.g., those requiring years of exposure). We explored the use of data mining on clinical notes to detect novel adverse drug effects. We constructed a knowledge base using UMLS and other data sources that could classify drug-finding pairs as currently known adverse effects (drug causes finding), known indications (drug treats/prevents finding), or unknown relationship. We used natural language processing (NLP) to extract current medications and clinical findings (including diseases) from 360,000 de-identified history and physical examination (H&P) notes. We identified 35,000 interesting co-occurrences of medication-finding concepts that exceeded threshold probabilities of appearance. These involved ~600 drugs and ~2000 findings. Among the identified pairs are several that the FDA recognized as harmful in postmarketing surveillance, including rofecoxib and heart attack, rofecoxib and stroke, statins and diabetes, and statins and memory loss. Our preliminary results illustrate both the problems and potential of using data mining of clinical notes for adverse drug effect discovery.

Pilot studies have demonstrated the feasibility of using mobile technologies as a platform for electronic patient-centered medication management. Such tools may be used to intercept drug interactions, stop unintentional medication overdoses, prevent improper scheduling of medications, and to gather real-time data about symptoms, outcomes, and activities of daily living. Unprompted text-message communication with patients using natural language could engage patients in their healthcare but presents unique natural language processing (NLP) challenges. A major technical challenge is to process text messages and output an unambiguous, computable format that can be used by a subsequent medication management system. NLP challenges unique to text message communication include common use of ad hoc abbreviations, acronyms, phonetic lingoes, improper auto-spell correction, and lack of formal punctuation. While models exist for text message normalization, including dictionary substitution and statistical machine translation approaches, we are not aware of any publications that describe an approach specific to patient text messages or to text messages in the domain of medicine. To allow two-way interaction with patients using mobile phone-based short message service (SMS) technology, we developed the Patient-centered Automated SMS Tagging Engine (PASTE). The PASTE webservice uses NLP methods, custom lexicons, and existing knowledge sources, to extract and tag medication concepts and action concepts from patient-generated text messages. A pilot evaluation of PASTE using 130 medication messages anonymously submitted by 16 volunteers established the feasibility of extracting medication information from patient-generated medication messages and suggested improvements. A subsequent evaluation study using 700 patient-generated text messages from 14 teens and 5 adults demonstrated improved performance from the pilot version of PASTE, with F-measures over 90% for medication concepts and medication action concepts when compared to manually tagged messages. We report on recall and precision of PASTE for extracting and tagging medication information from patient messages.

Achieving micron resolutions in magnetic resonance imaging is constrained first by limitations in available signal strength as voxel sizes decrease, and second, by acceptable acquisition times due to the large data sets required. The latter is problematic due to an increased sensitivity to patient bulk motion and physiological effects, and prevalence of distortion and blurring artifacts caused by susceptibility variation. Signal constraints can be mitigated using ultra-high field strengths, such as 7T, but face field dependent challenges such as increased B1 inhomogeneity and shorter T2* values. Scan times can be minimized using reduced field-of-view (FOV) imaging techniques that localize excitations to smaller regions of an object to achieve diminished imaging dimensions, but have largely been unexplored at 7T. To address this deficiency with the goal of improving human imaging resolutions, this thesis first implements and compares multiple reduced-FOV methods at 7T, assessing relative ability to localize excitation, suppress unwanted signal, minimize artifacts, and constrain power deposition. Inner-Volume Imaging (IVI) and Outer-Volume Suppression (OVS) methods optimized from this comparison are then synergistically combined with rapid parallel and echo planar imaging (EPI) techniques to obtain 160 to 500 μm2 in vivo images throughout the human brain in 3 to 12 minutes, accelerated 160 to 1400 fold for multi-slice and 3D scans, respectively. Compared to full-FOV scans, this approach demonstrates reduced geometric distortion and motion artifacts, with improved visibility of features at the high resolution. The parallel reduced-FOV method is similarly applied for diffusion tensor and cervical spine imaging prone to motion and susceptibility artifacts to obtain 1mm2 DTI images and 300 μm2 in the spine, with localized measurement of diffusion properties. Overall, the reduced-FOV approach provides reduction in scan times, artifact minimization, and achieves resolutions that exceed prior studies.

Use of Bayesian networks (BN) has increased in medicine. Traditionally, BNs have been developed by experts or from the current literature. Several applications implement "off the shelf" BN structure learning algorithms, but few implementations have been evaluated. We compared six "off the shelf" BN structure learning algorithms and an expert-developed BN using two years of data from a curated emergency department (ED) overcrowding database. We used ED ambulance diversion as the reference standard. Eighteen variables selected from a previous study were used for prediction. BN structures were learned from a data set for predicting ED diversion one hour in advance. The data set was discretized using equal frequency and equal width discretization. Each BN structure learning algorithm developed a structure based on each data set. We used area under the receiver operating characteristic curve (AUC), negative log likelihood, and Akaike information criterion to compare the structures as they predicted ED diversion at 1, 2, 4, 6, 8, and 12 hours in advance. Both the training and test data sets contained >100,000 data points. The ED was on ambulance diversion 22% of the time. The machine-learned networks were complex, with >3,000 conditional probabilities, compared to the expert-developed network, with 365. Both the best performing machine-learned structure and the expert-developed network had an AUC of 0.95 predicting diversion at one hour and 0.94 predicting diversion at two hours in advance. The machine-learned BN performed as well at the expert-developed BN. The expert-developed network was parsimonious, but required extensive user involvement.

Infrared neural stimulation (INS) has been shown to induce neural activity with spatial selectivity without inducing a stimulation artifact or necessitating tissue contact. Most experiments with this technology have been done in mammals, but much is still unknown about INS. Characterization of the experimentally tractable nervous system of the marine mollusk, Aplysia californica, will allow us to answer fundamental questions. In these studies, INS feasibility in Aplysia is shown and characterized with respect to repetition rate, pulse duration, wavelength and temperature. Nerve recordings were taken while stimulating buccal nerve 3 of the buccal ganglion using pulsed infrared light. At each parameter value, the nerve was stimulated at a random radiant exposure level (J/cm2), and observation of stimulation was reported. Stimulation thresholds were calculated as the effective dose (ED50) outputted by a probit regression. No change in stimulation thresholds were investigated at laser repetition rates ranging between 0.5-15 Hz. Stimulation at pulse durations ranging from 3-10 ms showed no change in threshold while it decreased below and increased above that range. Studies showed that a wavelength of 1.875 μm was more efficient for inducing action potentials than at 1.865 μm. Investigations of the ambient temperature challenged previous work by showing that at lower temperatures (0 Celsius) the threshold increased. Complex behavioral patterns were induced using INS in neural networks providing new directions for future clinical devices. Having shown feasibility in Aplysia, we believe this is a useful model for further studies on the physiological mechanism and optimal laser parameters of INS.

With the worlds increasing energy demands, there exists a tremendous need for the development and industrialization of energy-dense biofuels. Metabolically engineered cyanobacteria provide a promising means, as the utility of photosynthesis bypasses the need for harvesting, transporting and deconstructing biomass. The idea proposed is to manipulate circadian pathways in an effort to optimize cyanobacterial isobutyraldehyde (IBA) production. Bioluminescence has proven successful as a real-time reporter of circadian gene activity, so the objective was to determine whether it is a feasible means of measuring IBA levels. The automated high-throughput system developed expands upon the Kondotron system, utilizing a twelve-channel turntable with stepper motor and a CCD-cooled camera. The new system makes use of commercially available parts and is controlled entirely with custom LabView software. It features several software improvements, most notably to colony selection and processing. Luminescence activity was found to increase with increasing IBA vapor concentrations, thereby allowing for efficient screening and monitoring of IBA-producing mutants. It is expected that system performance will only improve when long-chain aldehydes, alcohols, or alkanes are the desired end-product. A similar high-throughput system could be developed for monitoring bacterial fluorescence.

Assessment of the restoration of load-bearing function is the central goal in the study of bone fracture healing. However, bone fracture calluses are inhomogeneous and irregular materials and this complexity has led to considerable uncertainty in the assessment of biomechanical property improvement or impairment during various therapeutic interventions and genetic models of pathological fracture healing. Unfortunately, as a result, arguably one of the most important criteria, mechanical stability, is the least resolved with respect to fracture healing assessment. To address this issue, an inverse finite element analysis (FEA) approach was developed in which biomechanical testing and microCT imaging are integrated through the use of computational modeling to determine mechanical properties of the healing fracture callus tissue. The presented work serves to evaluate the inverse analysis as a functional fracture healing assessment methodology in comparison to more traditional imaging and biomechanical testing measures within the context of normal fracture healing and a therapeutic system involving mesenchymal stem cell (MSC) transplantation. As compared to traditional fracture healing metrics, the results demonstrate that the inverse FEA approach: (1) was the only metric to successfully detect both longitudinal and therapeutic responses, and (2) performed significantly better at late-stage healing time points, where traditional metrics failed. The inverse analysis also added insight to the role of MSCs in fracture healing by demonstrating both accelerated healing and therapeutic benefit at late-stage healing. Additionally, a systems-based approach was developed for the generation of ease-of-use enhancements to the inverse analysis methodology in order to facilitate a wider usage among bone fracture biology groups whom are not experts in computational analysis. This was accomplished by the construction of an online web-enabled model submission system in which bone fracture callus microCT imaging and biomechanical testing data are collected with a minimal amount of pre-processing on a remote user node and submitted to a compute node which builds and executes the inverse model for material property reconstruction. In conclusion, the inverse FEA approach is shown to be a sensitive and functional fracture healing measure and provides a significant first-step towards normalizing the often challenging process of assessing mechanical integrity of healing fractures.

This PhD Dissertation project had as its objectives: (1) to develop MIASMA, a potentially open-source Medical Informatics Application for Systematic Microbiological Alerts that uses recently developed methods (e.g., from syndromic surveillance and from heuristic observations) to detect single-hospital outbreaks of both commonly occurring and rare bacterial, viral, and fungal species; (2) to deploy MIASMA in the Vanderbilt University Hospital (VUH) for use by the Department of Infection Control and Prevention; (3) to compare the alerting timeliness, positive predictive value, and sensitivity of MIASMA to current VUH infection control practices; and (4) to evaluate the utility of MIASMA when used to supplement current VUH infection control practices.