Carl Ehrett

Dr. Carl Ehrett

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 March 2025 in PNAS Nexus
Verified
Journal Article
Open access
MWMorgan Wack
Morgan Wack
CECarl Ehrett
Carl Ehrett
DLDarren Linvill
Darren Linvill
PWPatrick Warren
Patrick Warren
Can AI bolster state-backed propaganda campaigns, in practice? Growing use of AI and large language models has drawn attention to the potential for accompanying tools to be used by malevolent actors. Though recent laboratory and experimental evidence has substantiated these concerns in principle, the usefulness of AI tools in the production of propaganda campaigns has remained difficult to ascertain. Drawing on the adoption of generative-AI techniques by a state-affiliated propaganda site with ties to Russia, we test whether AI adoption enabled the website to amplify and enhance its production of disinformation. First, we find that the use of generative-AI tools facilitated the outlet’s generation of larger quantities of disinformation. Second, we find that use of generative-AI coincided with shifts in the volume and breadth of published content. Finally, drawing on a survey experiment comparing perceptions of articles produced prior to and following the adoption of AI tools, we show that the AI-assisted articles maintained their persuasiveness in the postadoption period. Our results illustrate how generative-AI tools have already begun to alter the size and scope of state-backed propaganda campaigns.

19 November 2024 in JMIR Medical Education
Verified
Journal Article
Open access
CECarl Ehrett
Carl Ehrett
SHSudeep Hegde
Sudeep Hegde
KAKwame Andre
Kwame Andre
DLDixizi Liu
Dixizi Liu
TWTimothy Wilson
Timothy Wilson
Background: Generative large language models (LLMs) have the potential to revolutionize medical education by generating tailored learning materials, enhancing teaching efficiency, and improving learner engagement. However, the application of LLMs in health care settings, particularly for augmenting small datasets in text classification tasks, remains underexplored, particularly for cost- and privacy-conscious applications that do not permit the use of third-party services such as OpenAI’s ChatGPT. Objective: This study aims to explore the use of open-source LLMs, such as Large Language Model Meta AI (LLaMA) and Alpaca models, for data augmentation in a specific text classification task related to hospital staff surveys. Methods: The surveys were designed to elicit narratives of everyday adaptation by frontline radiology staff during the initial phase of the COVID-19 pandemic. A 2-step process of data augmentation and text classification was conducted. The study generated synthetic data similar to the survey reports using 4 generative LLMs for data augmentation. A different set of 3 classifier LLMs was then used to classify the augmented text for thematic categories. The study evaluated performance on the classification task. Results: The overall best-performing combination of LLMs, temperature, classifier, and number of synthetic data cases is via augmentation with LLaMA 7B at temperature 0.7 with 100 augments, using Robustly Optimized BERT Pretraining Approach (RoBERTa) for the classification task, achieving an average area under the receiver operating characteristic (AUC) curve of 0.87 (SD 0.02; ie, 1 SD). The results demonstrate that open-source LLMs can enhance text classifiers’ performance for small datasets in health care contexts, providing promising pathways for improving medical education processes and patient care practices. Conclusions: The study demonstrates the value of data augmentation with open-source LLMs, highlights the importance of privacy and ethical considerations when using LLMs, and suggests future directions for research in this field.

30 April 2024 in Sensors
Verified
Journal Article
Open access
PDF file
NDNeelima Dahal
Neelima Dahal
CPCaroline Peak
Caroline Peak
CECarl Ehrett
Carl Ehrett
JOJeffrey Osterberg
Jeffrey Osterberg
MCMin Cao
Min Cao
PWPingshan Wang
Pingshan Wang
Label-free measurement and analysis of single bacterial cells are essential for food safety monitoring and microbial disease diagnosis. We report a microwave flow cytometric sensor with a microstrip sensing device with reduced channel height for bacterial cell measurement. Escherichia coli B and Escherichia coli K-12 were measured with the sensor at frequencies between 500 MHz and 8 GHz. The results show microwave properties of E. coli cells are frequency-dependent. A LightGBM model was developed to classify cell types at a high accuracy of 0.96 at 1 GHz. Thus, the sensor provides a promising label-free method to rapidly detect and differentiate bacterial cells. Nevertheless, the method needs to be further developed by comprehensively measuring different types of cells and demonstrating accurate cell classification with improved machine-learning techniques.

1 March 2021 in Journal of Verification, Validation and Uncertainty Quantification
Verified
Journal Article
Open access
PDF file
CECarl Ehrett
Carl Ehrett
DBD. Andrew Brown
D. Andrew Brown
CKChristopher Kitchens
Christopher Kitchens
XXXinyue Xu
Xinyue Xu
RPRoland Platz
Roland Platz
SASez Atamturktur
Sez Atamturktur
Calibration of computer models and the use of those design models are two activities traditionally carried out separately. This paper generalizes existing Bayesian inverse analysis approaches for computer model calibration to present a methodology combining calibration and design in a unified Bayesian framework. This provides a computationally efficient means to undertake both tasks while quantifying all relevant sources of uncertainty. Specifically, compared with the traditional approach of design using parameter estimates from previously completed model calibration, this generalized framework inherently includes uncertainty from the calibration process in the design procedure. We demonstrate our approach to the design of a vibration isolation system. We also demonstrate how, when adaptive sampling of the phenomenon of interest is possible, the proposed framework may select new sampling locations using both available real observations and the computer model. This is especially useful when a misspecified model fails to reflect that the calibration parameter is functionally dependent upon the design inputs to be optimized.

24 March 2025 in IEEE Sensors Journal
Verified
Journal Article
Closed access
CECarl Ehrett
Carl Ehrett
MKMatthew Keagle
Matthew Keagle
BBBenjamin Braun
Benjamin Braun
NHNitya Harikumar
Nitya Harikumar
JOJeffrey Osterberg
Jeffrey Osterberg
PWPingshan Wang
Pingshan Wang
Microwave sensing of biological cells is of great interest for its noncontact, noninvasive nature and high-sensitivity, high-resolution potential. Nevertheless, the lack of measurement specificity has been a critical problem. In this work, we show that highly reproducible, unbiased microwave data of single cells, analyzed by classical machine learning algorithms, enable effective cell classification. Yeast cells, including Saccharomyces cerevisiae, Saccharomyces pastorianus, Candida albicans, Candida tropicalis, Candida parapsilosis, and Candida krusei, were measured with microwave flow cytometers at frequencies between 0.265GHz and 7.65GHz. Four sets of yeast cell data with 7,941, 310, 130, and 200 cells were obtained. Machine learning models, including Random Forest, XGBoost, K-Neighbors, Decision Tree, and Support Vector Machine, were trained to classify cell types and viability. Accuracies of up to 86.3% were obtained for cell type classification. The results show that the addition of computer vision features via cell imaging helps improve accuracy, and that ML model performance depends on its targeted cell traits, the number of frequencies applied, and the values of chosen frequencies. Standardized data acquisition and significantly more cell data are needed for further ML classification investigation.

26 July 2026
Verified
Conference Paper
Open access
CECarl Ehrett
Carl Ehrett
AFAlex Feltus
Alex Feltus
DDDouglas Dawson
Douglas Dawson
ZGZachary Gerstner
Zachary Gerstner
TCTzu-Yu Chu
Tzu-Yu Chu
Agentic AI assistants show promise for accelerating scientific discovery, but production high-performance computing (HPC) deployment adds requirements such as queued execution, resource governance, and auditable provenance. We present BioCodex (palmetto-mcp), an HPC-native AI tooling framework that enables an assistant to discover scientific tools, generate deterministic execution plans, and dispatch heavy computation asynchronously to a Slurm cluster. BioCodex separates planning from execution using two Model Context Protocol (MCP) servers: a bio server that validates arguments and emits hash-verified RunSpec artifacts, and a Slurm server that submits RunSpecs as batch jobs, polls status, and fetches standardized results and logs. Deployed on Clemson’s Palmetto cluster, BioCodex adds only small control-plane overheads, rejects tampered execution plans before scheduling, and supports representative end-to-end genomics workflows. These results suggest that deterministic planning artifacts plus scheduler-native asynchronous tools provide a practical pattern for embedding agentic assistants into production HPC environments.