<link rel="stylesheet" href="styles.cf390c6bfbe7555f.css">
Repository logo

The ScholarShip is a digital archive for the scholarly output of the ECU community. Its mission is to capture, preserve, and make available the intellectual output of East Carolina University's faculty, staff, and students.


Submit your work to the ScholarShip

How to submit work to the ScholarShip

Scholarly Communication Services

Communities & Collections in the ScholarShip

Select a community to browse its collections.

Recent Submissions

  • Item type:Item, Access status: Embargo ,
    Collagen, Interrupted: Synthesis and Application of Fluorescent Collagen Capsules
    (East Carolina University, 2026-07) Stanley, Jordan Lynne
    The separation of enantiomers in racemic mixtures remains a significant challenge in the pharmaceutical industry, as individual enantiomers of a chiral drug often exhibit markedly different biological activities, pharmacokinetics, and toxicity profiles. Current chiral separation methods, while effective, are frequently costly, time-intensive, and reliant on specialized instrumentation. The objective of this research is to develop a cost-effective, accessible, and reusable strategy for chiral separation through the use of modified collagen mimetic peptide (CMP) systems capable of selective molecular encapsulation. To this end, a series of novel naphthalimide backbone amino acids were synthesized and incorporated into various CMPs that follow the generic sequence Ac‑(POG)n‑Naphthalimide‑(POG)n‑GKK‑NH2. The naphthalimide residues function as both fluorescent probes and structural “interrupters” capable of forming a capsule-like cavity within a folded triple helix for the purpose of drug encapsulation and enantioselective recognition, enabling simultaneous investigation of peptide folding and intermolecular interactions. Circular dichroism studies demonstrate that incorporation of naphthalimide residues does not preclude triple helix formation, provided that the (POG)n sequences flanking the naphthalimide contain five or more Pro-Hyp-Gly repeat units. Thermodynamic analysis reveals a balance between enthalpic stabilization from interstrand hydrogen bonding and entropic contributions from hydrophobic interactions introduced by the naphthalimide moieties. Time-resolved fluorescence studies show that folding proceeds with first-order kinetics, consistent with a propagation-limited mechanism, and that naphthalimide clustering during helix formation results in measurable self-quenching. Circular dichroism and fluorescence studies with symmetrical CMP, Ac‑(POG)5‑Lnap‑(POG)5‑GKK‑NH2 (5‑Lnap‑5), reveal its ability to form a continuous triple helix through the naphthalimide and generate a capsule-like internal environment. Encapsulation studies demonstrate the ability of 5‑Lnap‑5 to host small molecules, while chiral experiments with R- and S-naproxen reveal enantioselective recognition, manifested through differences in folding kinetics and post-folding structural dynamics between enantiomers. Collectively, this work establishes naphthalimide-functionalized CMPs as structurally robust, reusable, and chiral-responsive host systems, providing a proof-of-concept platform for enantioselective encapsulation and separation technologies with potential applications in drug development, molecular recognition, and advanced biomaterials.
  • Item type:Item, Access status: Open Access ,
    IMPROVING INDOOR ENVIRONMENTS WITH REAL-TIME SENSOR DATA FOR BUILDING THERMAL COMFORT AND MOLD GROWTH
    (East Carolina University, 2026-12) Nguyen, Ngoc Thi Nhu
    Indoor environmental quality in office buildings is influenced by spatial and temporal variations in temperature, relative humidity, airflow, and heat transfer. Point sensors capture changes over time but may not identify localized conditions near diffusers, return grilles, perimeter surfaces, corners, and occupied zones. This study combined long-term environmental sensing, mold-risk analysis, and Computational Fluid Dynamics (CFD) to evaluate thermal comfort and mold growth risk in a pilot office building. Approximately two years of temperature and relative-humidity data were evaluated using dew point, equilibrium moisture content, threshold-based mold classification, the Conserv Mold Risk Score, and a days-to-mold-based Mold Risk Factor. Autodesk Revit and Autodesk CFD were used to model a branched corridor and an enclosed office room under representative summer and winter conditions. The simulation results revealed that the corridor developed high-velocity regions near mechanical terminals and weakly mixed zones at ends, corners, branches, and intersections. In the office room, a localized near-floor layer between approximately 69.04°F and 72.40°F was identified within an otherwise comparatively uniform bulk-air region. Available room results produced Predicted Mean Vote (PMV) values from approximately -0.37 to +0.16, indicating generally neutral to slightly cool conditions with localized discomfort near the floor and supply-influenced regions. Sensor analysis also identified recurring humidity-sensitive conditions in above-ceiling, perimeter, and other low-circulation locations. The findings show that acceptable average conditions can coexist with localized comfort and moisture concerns. The integrated framework supports targeted HVAC adjustments and strategic sensor placement at multiple heights and in stagnation, perimeter, and transition zones.
  • Item type:Item, Access status: Open Access ,
    A Graph-Augmented AI System for Automated Changelog Generation in Secure Environments
    (East Carolina University, 2026-12) Richmond, Eli
    Secure software engineering environments need AI assistance without exposing proprietary source code to external services. Changelog generation is a useful test case because it requires grounded translation from repository changes to user-facing technical communication. A graph-augmented AI system is presented for automated changelog generation in secure environments. The system builds a directed repository graph, retrieves release-relevant context, and prompts an LLM to generate Markdown changelogs. When a release range exceeds the prompt budget, the system summarizes diff chunks and merges intermediate summaries before final generation. The completed evaluation adds an LLM judge that scores semantic quality, factual alignment, coverage, usefulness, and whether the generated changelog is good enough to stand beside the reference. The evaluation compares multiple repository/model runs and includes a separate context-window experiment on selected releases to study how compression affects output quality. These results support the end-to-end graph-aware pipeline as a practical time-saving framework for secure changelog assistance. The system does not eliminate maintainer review, but it produces a strong first draft far faster than manual changelog construction. The code and result artifacts are available at https://github.com/erichmond33/ secure-changelog-generation.
  • Item type:Item, Access status: Open Access ,
    Cardiac Physiological Dynamics in Implantable Loop Recorder Patients: A Bayesian M-MELS Model Analysis of Physical Activity Trends
    (East Carolina University, 2026-07) Sargeant, Maeve M.
    Physical activity (PA) is consistently low among individuals with cardiovascular disease, yet average activity summaries do not explain how activity unfolds in relation to daily physiological conditions. This limitation is especially relevant for patients with implantable loop recorders (ILRs), whose activity occurs in the context of ongoing cardiac monitoring, ambiguous symptoms, and day-to-day physiological fluctuation. The present study examined whether daily PA in ILR patients is characterized by structured temporal patterns when modeled alongside heart rate variability (HRV) and mean ventricular rate (MVR). The analytic sample included 73 ECU Health Cardiology patients with Biotronik ILRs and at least one year of device-recorded data. Across the monitoring period, patients contributed 72,597 daily observations. Daily device-derived PA, HRV, and MVR were modeled using a Bayesian multivariate mixed-effects location-scale vector autoregressive model. This approach estimated person-specific average levels, residual day-to-day variability, autoregressive persistence, and cross-lagged associations among PA, HRV, and MVR. Person-level atrial fibrillation (AF) burden and premature ventricular contraction (PVC) burden were included as between-person predictors of average levels, variability, and temporal dynamics. Patients showed meaningful between-person differences in typical PA, HRV, and MVR, as well as residual day-to-day variability after accounting for lagged structure. Intraclass correlations indicated that 67% to 76% of variance in the three primary signals reflected between-person differences, while each signal retained meaningful within-person daily fluctuation. Higher average PA was strongly associated with greater PA innovation variability, indicating that individuals with higher typical activity also showed broader residual day-to-day fluctuation in activity. All three variables demonstrated significant autoregressive persistence, with MVR showing the strongest carryover across days. Several cross-lagged associations were also observed: higher prior-day MVR predicted lower next-day PA, higher prior-day PA predicted higher next-day HRV and lower next-day MVR, and HRV and MVR showed reciprocal negative cross-lagged associations. In contrast, prior-day HRV did not meaningfully predict next-day PA. AF burden was not credibly associated with average levels, variability, or temporal dynamics of PA, HRV, or MVR. PVC burden was not associated with average levels or most dynamic parameters, but higher PVC burden was associated with greater HRV innovation variability. These findings indicate that daily ILR data contain temporal structure that is not captured by static activity summaries. PA in ILR patients appears to be embedded within a dynamic physiological context, rather than functioning solely as a marker of motivation, adherence, or disease burden. Modeling activity together with HRV and MVR may help distinguish patient-specific patterns of behavioral engagement, physiological fluctuation, and temporal coupling. Future work should integrate device-derived dynamics with symptom reports, fear of exertion, clinical indication, medication use, and functional outcomes to clarify which behavioral-physiological patterns correspond to clinically meaningful targets for intervention.
  • Item type:Item, Access status: Embargo ,
    Thoracoabdominal Pressure Coupling During Prone Versus Supine Cardiopulmonary Resuscitation: A Cadaver-Based Mechanistic Study
    (East Carolina University, 2026-12) Smith, Zachary
    Cardiac arrest in the prone position is uncommon but requires a time-critical response in perioperative and critical care settings. Turning a patient supine can require several personnel to manage the airway, protect vascular access, position equipment, and maintain surgical exposure. Current guidelines allow consideration of prone cardiopulmonary resuscitation (CPR) when safe and when immediately turning the patient to the supine position is not feasible. Clinical reports and limited experimental studies have shown that prone compressions can generate arterial pressures comparable with, or greater than, those observed during supine CPR. The relative contributions of abdominal and thoracic pressure to these arterial responses remain unresolved. This dissertation examined how pressure is transmitted across the abdominal, thoracic, and central arterial compartments during CPR in the prone and supine positions. An integrative review examined cardiac pump, thoracic pump, and abdominal pump contributions and identified a lack of simultaneous intra-abdominal, intrathoracic, and arterial pressure measurements. A prospective within-specimen crossover study was conducted in two thawed fresh-frozen adult human cadavers. Each cadaver underwent three 2-minute trials of standardized mechanical compressions in both prone and supine positions in a counterbalanced sequence. Solid-state pressure catheters recorded pleural, peritoneal, and central arterial pressure synchronously. Human cadavers preserved the gross anatomic relationships needed to examine pressure transmission and permitted multi-compartment instrumentation. This would be impractical and ethically inappropriate during active resuscitation of a living patient. Prone CPR increased peak pleural, peritoneal, and arterial pressures by 7.04, 21.69, and 15.40 mmHg, respectively. Mean peritoneal and arterial pressures increased by 16.22 and 9.90 mmHg. The mean peritoneal-minus-pleural pressure difference shifted from −8.46 mmHg during supine CPR to 4.85 mmHg during prone CPR. Pleural pressure-time area increased from 1.62 to 5.46 mmHg·s, and arterial pressure-time area increased from 2.96 to 6.42 mmHg·s. In a secondary exploratory experiment with separate supine control trials, manual abdominal pressure increased mean pleural, peritoneal, and arterial pressures by 8.19, 17.29, and 6.30 mmHg. Circumferential binding increased the corresponding means by 6.09, 13.12, and 8.96 mmHg. Manual pressure produced higher pleural mean and cycle-maximum pressures than binding, whereas binding produced higher arterial mean and cycle-minimum pressures with similar arterial cycle maxima. Together, the findings support an abdominal contribution to compression-generated thoracic and arterial pressure. Thoracoabdominal coupling is a proposed mechanism in which abdominal pressure and restraint modify the accommodation of chest compression through the diaphragm and abdominal wall. Direct measurement of diaphragm motion, blood flow, and perfusion pressure gradients is needed to test this explanation. The dissertation provides a reproducible human cadaver method and a testable framework linking positioning and abdominal interventions to pressure generation, informing subsequent mechanistic studies.