Annual Research Symposium

Follow

• 

  AI-SCM CMM: A Capability Maturity Model for Artificial Intelligence Integration in Supply Chain Management

  Lordt Becklines, Omar F. El-Gayar, Patti Brooks, and Insu Park

  Artificial intelligence is increasingly deployed in supply chain management, yet many organizations struggle to align adoption efforts with process readiness, data quality, governance, and workforce capabilities, and they still lack validated supply chain specific roadmap for assessing readiness, sequencing investments, and reducing implementation risk. This study develops and evaluates a Capability Maturity Model for Artificial Intelligence Integration in Supply Chain Management to address that gap. Using a design science research approach, the study synthesizes prior literature and practitioner knowledge to define maturity dimensions, capability indicators, and staged progression levels for AI integration in supply chain contexts. The artifact and assessment instrument were iteratively refined and validated through expert review using a Delphi based process to strengthen relevance, clarity, and practical usability. The resulting model enables organizations to assess current capability, identify priority gaps, and plan improvement actions aligned with operational goals across planning, design, and execution activities. A case-based application demonstrates how the instrument produces an organizational maturity profile and supports decision making on capability development priorities. This research contributes a practitioner ready assessment and planning tool and advances scholarship by offering a validated framework for staged and performance aligned AI integration in supply chain management.

• 

  Building Explainable RAG-based Clinical Decision Support

  Irina Pecherskaia, Jason Mixon, Andrew Behrens, and Andrew Smith

  The objective of this research is to demonstrate how theoretical Explainable AI (XAI) principles are operationalized into a functional, accurate, and trustworthy Retrieval-Augmented Generation (RAG) prototype for assisting healthcare professionals in Clinical Decision Support. While Large Language Models (LLMs) are powerful in knowledge synthesis, they often lack transparency and produce hallucinations. RAG can help address the problem by mitigating LLM hallucinations by grounding outputs in verified medical literature and clinical guidelines. This project bridges the clinical trust gap by implementing five core design principles into a RAG prototype that grounds AI responses in verified medical literature.

• 

  Comparative Analysis of Shipping Costs Across Multiple Routes Using FedEx API Data

  Valerija Curikova and Andrew Kramer

  Shipping costs are an important factor for both individuals and companies, especially in industries like healthcare where medical material should be delivered quickly and reliably. While working as an intern at Avera Health in the Supply Chain logistics team, I was introduced to the challenge of the high cost of delivering medical supplies to different patient locations. Shipping prices vary depending on distance, route direction, and service type. Using real-time data from the FedEx API allows for more accurate analysis of shipping costs and helps identify opportunities to optimize delivery decisions and reduce logistics expenses.

• 

  Constructing Information Systems: Technology Frames and Career Interest Among Middle School Students

  Sai Mounika Chintalapudi and Cherie Noteboom

  Information Systems (IS) plays a critical role in modern society, supporting organizations, healthcare systems, businesses, and digital infrastructure. Despite its importance, many students develop limited or inaccurate perceptions of the IS field at an early age, which may influence their willingness to pursue IS-related careers. Middle school is an important stage where students begin forming career interests and professional identities. However, little research has examined how younger students interpret and understand the field of Information Systems. Technology Frames of Reference (TFR) explains how individuals interpret technology through cognitive frames related to its nature, purpose, and use. While TFR has traditionally been applied in organizational contexts, this study adapts the framework to examine how middle school students construct meaning about Information Systems and how these interpretations relate to career interest.

• 

  Evaluating the Impact of AI and Blockchain for Supply Chain Risk Mitigation: A Predictive Analytics Approach

  Sai Neelima Seru and David Zeng

  The increasing adoption of artificial intelligence (AI) in supply chain operations is transforming how organizations detect, govern, and respond to risk, raising important questions regarding digital trust, accountability, and transparency. AI models enable predictive risk assessment using large-scale logistics data. However, their outputs are often difficult to audit or independently verify. In contrast, blockchain technology provides immutable and tamper-evident records but lacks predictive capabilities. This study empirically evaluates an integrated AI-Blockchain environment using a real-world e-commerce logistics dataset (Olist Brazilian dataset, September 2016-October 2018). The analysis compares AI-only, blockchain-only, and integrated configurations across both predictive and governance dimensions. Results indicate that the integrated system improves risk detection coverage compared with AI alone, while blockchain adds governance capabilities, including tamper detection and end-to-end event traceability. The findings demonstrate that AI and blockchain address different operational failure modes, and their integration creates governance capabilities that neither technology can provide independently.

• 

  Evaluating the Reliability and Equity Implication of Acute Hospital Readmission Metrics in New Zealand

  Christopher D. Elce, Martinson Ofori, and Andrew Behrens

  Acute 28-day hospital readmission rates are widely used in New Zealand to monitor hospital quality and health system performance. However, the policy value of readmission metrics depends on whether observed variation reflects real clinical differences or is driven by data-quality artifacts and population structure. This study evaluates the reliability of published benchmarked readmission rates for all Districts of Service.

• 

  EXAIM: A Real-Time Explainability Middleware for Multi-Agent Clinical Decision Support Systems

  Abem Woldesenbet and Andrew Behrens

  PROBLEM:

  • Current LLM-based CDSS only provide post-hoc explanations, not live transparency.
  • Clinicians cannot observe diagnostic reasoning as it unfolds.
  • Multi-agent reasoning generates long, overwhelming token streams.

  GAP:

  • No existing CDSS supports token-level, real-time explainability.
  • Flow control and semantic interpretation are tightly coupled in current systems.
  • Prior work focuses on static summaries or rule-based rationales.

  PURPOSE:

  • Objective: Design EXAID, a middleware that summarizes evolving diagnostic reasoning as tokens are streamed.
  • Goal: Enable clinicians to monitor live logic without being overwhelmed by raw data.

• 

  Hardware Side-Channel Security of Quantum System Controllers: A Timing Attack Perspective

  Darpan Basnet, Anshu Bista, and Varghese Vaidyan

  Quantum computers use classical embedded processors to sequence control pulses. On STM32-class microcontrollers, firmware emits operation classes (gates (X, Y, Z), measurements (MEAS), and timing barriers (WAIT)) at precise intervals. If execution time depends on a secret value, an attacker with a logic analyzer can recover that secret. This classical control plane is a largely overlooked attack surface. This work presents a simulation-based investigation of timing side-channel leakage in quantum-control firmware sequencers.

• 

  Interactions of Bad Actors with Honeypots

  Maryam Aliyeva and Andrew Kramer

  Throughout daily life, computer users encounter different forms of malware, such as ransomware, adware, viruses, trojan horses, and spyware. To protect networks, individuals need to be able to comprehensively analyze a malicious actor’s behavior and tactics. These observations can be made through the usage of honeypots – a computer security mechanism set to track and deflect unauthorized activity. Honeypots frequently operate like decoys of legitimate websites that perfectly mimic an existing database, which makes them a valuable tool for researching cyber criminals’ behavior.

• 

  LLM Security Agents: Harness Design and Static vs Dynamic Challenges

  Joe Hammond, Eddie French, and Austin O'Brien

  LLM security agent performance depends on two factors: harness design and challenge type. A poorly designed harness prevents models from recovering from failures, while challenge type determines baseline difficulty. We tested these factors across two experiments. In Experiment 1, we evaluated 6 harness-model combinations against 5 live HackTheBox machines requiring scanning, enumeration, exploitation, and privilege escalation across SSH, SMB, FTP, HTTP, and DNS. In Experiment 2, we benchmarked 10 frontier models via Claude Code Router on 5 challenges from Cybench spanning pwn, forensics, web, reverse, and crypto categories using a Pass@3 metric. Our experiments show models achieving 100% with one harness but scoring 0% with another, and the same model solves ~90% of static challenges but only ~20% of dynamic ones. The key insight is that dynamic challenges are solvable when the harness enables both efficient routine operations and a failure-recovery loop.

• 

  Mid-Generation Jailbreaks in Open-Source LLMs Using a Pause-and-Edit Attack

  Aman Singh, Komal More, Samyam Aryal, and Mark Spanier

  Large Language Models (LLMs) are widely used in AI assistants, chatbots, and decision-support systems. To prevent harmful responses, most LLMs rely on safety alignment mechanisms that generate refusal responses when users request unsafe content. However, most safety evaluations assume that alignment is only required at the start of generation. In this research, we investigate a mid-generation jailbreak attack called Pause-and-Edit, where a refusal response is interrupted, modified, and resumed. This manipulation can cause the model to override its original safety decision and generate harmful instructions. Our study evaluates how vulnerable modern open-source LLMs are to this type of attack.

• 

  Packet Scheduling in Mixed Traffic Networks: A Simulation Study Using NS-3

  Landon Mohr

  Packet scheduling determines the order in which packets are transmitted when multiple flows compete for a network link. Schedulers directly affect key performance metrics including latency, jitter, packet loss, and throughput. While scheduling algorithms have been widely studied, fewer evaluations examine their behavior under mixed traffic workloads representative of real networks. This study uses the NS-3 network simulator to evaluate how different packet scheduling algorithms behave under congestion across several representative traffic profiles.

• 

  Postmortem Analysis of Israel's 7 October 2023 Intelligence Failure

  Emily Helgeson and William Bendix

  On October 7, 2023, Hamas launched an attack from the Gaza Strip into Israel, resulting in the deaths of 1,200 people and triggering a larger conflict. Despite possessing impressive intelligence capabilities, Israel failed to anticipate the attack. This project provides one of the first postmortem analyses of the October 7 intelligence failure and demonstrates methods for assessing the role and significance of key contributing factors.

• 

  Post-Quantum Cryptography Secure Communication, IoT, and Blockchain

  Nidhish Bhanse and Mark Spanier

  Modern public-key cryptography, such as RSA and Elliptic Curve Cryptography (ECC), plays a crucial role in securing data. However, the development of quantum computing threatens the security of data encrypted with these methods. Data encrypted today might be decrypted in the future due to the increased power of quantum computers. To combat this, the National Institute of Standards and Technology (NIST) has developed new standards for post-quantum cryptography. The following research aims to provide an analysis of these NIST post-quantum cryptographic algorithms and their potential for use in various secure communication protocols and platforms.

• 

  Secure-Home: Detect and Redact PII

  Hajar Niroomand and Omar F. El-Gayar

  Personally Identifiable Information (PII) leakage from home environments poses significant identity theft risks. While enterprise networks employ robust security measures, firewalls, intrusion detection systems, and access controls, these protections rarely extend to home settings, creating a critical security gap. Current firewall technologies lack the capability to detect and scrub PII from outbound traffic, leaving vulnerable populations such as children, elderly users, and remote workers exposed. This design science research proposes Secure-Home, a prototype outbound inspection tool that detects and redacts clear-text PII before data leaves home networks. Using rule-based detection and selective redaction, Secure-Home provides a practical last line of defense tailored to domestic environments. Our research addresses two key questions: (1) Can outbound home network traffic be effectively monitored to detect PII leakage? Can real-time PII scrubbing reduce identity theft risks without disrupting household usability? The significance of this research is threefold. First, it addresses a critical vulnerability in an increasingly connected home environment where smart devices expand the attack surface. Second, it protects overlooked populations (children, elderly users, and families) who lack enterprise-grade security. Third, it contributes empirical evidence and a novel framework for home network security protocols. This project will deliver measurable reductions in PII leakage, provide peace of mind for families, and establish empirical data on tool effectiveness to guide future security protocols for non-enterprise settings. By bridging the gap between enterprise security and home protection, Secure-Home offers an innovative approach to safeguarding personal data where vulnerability is highest.

• 

  Streamlining Literacy Assessment

  Sheila Mulder, Katie Anderson, and Samuel W. Flint

  Reading proficiency scores are low throughout the nation. Reading proficiency is a foundational skill that supports academic success across all content areas. Skilled reading involves multiple interrelated components. Assessment can support teachers to identify skill deficits and inform instruction; however, rural teachers lack the time, resources, and specialized support to translate the data into effective instruction and interventions. To better contextualize this need, we received 59 surveys from South Dakota teachers with 23 of the surveys being valid responses, finding that teachers feel prepared to teach reading and struggling readers. Teachers reported that they spend relatively little time in assessment, and they frequently make data-backed decisions. However, on the assessment knowledge survey questions, respondents showed low actual preparedness to use assessment data based on their responses. To help teachers translate the assessment data they have, we are developing and testing a data-driven tool to aid teachers to efficiently identify student skill deficits, through research-backed selection of diagnostics. From this diagnostic data our tool groups and suggests efficient, relevant, evidence-based interventions. Ultimately with the goal that rural teachers receive the critical support and structure to effectively use reading assessment data to increase student literacy outcomes.

• 

  Using a Spherical Speaker to Explore Sound

  David Provance

  The goal of this project is to explore the acoustics of a space using a consistent, replicable design. To do this, I used an omnidirectional speaker to play a recording designed to study room tone. The idea comes from the benefits of real-world sounds and digital consistency. Using real-world sounds often provides a more accurate representation of the room; however, each sound is unique due to inconsistency. Normal speakers have directional output, which makes the reflections in a room less realistic. Using a spherical speaker, I could create sounds that output all around the source and are consistent from one recording to the next.

• 

  AI and Radiology: Exploring People, Process, and Technology through a Meta-Analysis

  Cherie Noteboom, Sai Mounika Chintalapudi, and Vahini Atluri

  Artificial Intelligence (AI) is increasingly being integrated into radiologists' workflows, enhancing their ability to make effective, data-driven decisions for patient care. This PRISMA-based meta-analysis examines systematic literature reviews (SLRs) to identify and expand upon the existing body of research regarding the types and roles of AI utilized in radiology. Covering academic databases from January 2019 to December 2024, the study employs a conceptual model based on the dimensions of People, Process, and Technology (PPT) to categorize the findings.

  The analysis highlights the evolving applications of AI in supporting radiologists' work and the specific roles benefiting from AI integration. This research provides valuable insights for practitioners and Information Systems (IS) researchers by identifying key trends and gaps in the literature. The study concludes with recommendations to enhance the understanding, adoption, and successful integration of AI into the complex professional responsibilities of radiologists. Initial findings suggest recommendations for improving the support of integration of AI into the categories of people, processes, and technology.

• 

  China vs Democracy: The Strategic Use of Malign Cyber Influence Campaigns

  Alexander Deak

  How and why do China’s methods for malign influence vary between different democracies?

• 

  Does the Use of Audits Decrease the Infection Rate in a Medical Care Setting?

  Kylie Borchert and Kristel K. Bakker

  Nosocomial infections are a significant health concern in medical settings. Reports of low compliance rates with hand hygiene standards, guidelines of which are outlined and mandated by the CDC and WHO, are frequent. Factors contributing to nonadherence include lack of knowledge and an unclear understanding of correct techniques (4). Evidence shows that improved hand hygiene can reduce infection rates (1), especially when healthcare providers are included in interventions that aim to improve compliance (4).

• 

  Ethanol Concentration in Gasoline

  Hannah Feser, Wyatt Olson, and Jeffrey Elbert

  For this research project, we will be measuring the concentration of ethanol present in different types of gasoline from various gas stations. The purpose of this project is to analyze and determine whether there is a substantial variance in the concentration, especially if it is possibly detrimental towards gas mileage and the product you intend to purchase is not what you are receiving. Our results have potential to lead to further research and discovery dependent upon if there is a significant variance found, this could mean a fault in the production or delivery lines of the gasolines.

• 

  If You Have Nothing to Hide, You Still Have Something to Fear: How Libraries Can Support Alternative Information Channels

  Abbie Steuhm

  Libraries in the U.S. have seen massive increases in book challenges in recent years. Since books are one of the library’s vital resources, censoring books can have a great impact on libraries and patrons. To counteract the effects of censorship, libraries need to adopt and support alternative information channels so information can still be accessed through other channels even if one is shut down.

• 

  Integrating Artificial Intelligence into Radiology: A Meta-Analysis of Educational Interventions and Technological Trends

  Cherie Noteboom, Vahini Atluri, and Sai Mounika Chintalapudi

  Radiology, a critical field in medical diagnostics, is undergoing a profound transformation as Artificial Intelligence (AI) is integrated into radiologists' workflows. This shift is redefining the roles and responsibilities of radiologists, professionals traditionally recognized for their expertise, autonomy, and societal prestige. As AI becomes increasingly embedded in radiological practice, effective educational interventions are essential to facilitate a seamless transition and optimize technology adoption. This study employs a PRISMA-based systematic literature review (SLRs), to examine AI applications in radiology and the educational strategies designed to support their implementation. Analyzing academic publications from January 2019 to December 2024, it categorizes AI technologies, theories, evolving radiologist roles and the instructional approaches used to enhance AI proficiency. The findings highlight the expanding role of AI in radiology and underscore the necessity of targeted education to bridge the gap between technological advancements and clinical practice. This research provides valuable insights for practitioners and information systems (IS) researchers by identifying key trends and gaps in existing literature. It concludes with recommendations for designing educational interventions that align with radiologists' evolving professional needs, ensuring the effective integration of AI into their Workflows.

• 

  Investigating the energetics and feeding ecology of a range of Azhdarchid Pterosaurs

  Kierra Miller and T. Alexander Deccechi

  Here we aim to gain insight into the ecology of Azhdarchid pterosaurs. By exploring their genetic requirements and feeding capabilities, we gain a deeper understanding of what their day-to-day life may have looked like.

• 

  Is Al-Driven threat detection an effective substitute for current threat detection architectures?

  Connor J. Ford

  This research evaluates the use of Artificial Intelligence (Al] in the development of cyber defense systems.