Collection

2025 Northeast Section Conference

Authors

Mohammad Rafiq Muqri; Simon Obeid; Gusteau Duclos, DeVry University

assessment system that evaluates the role of generative AI inproblem-solving. Machine Learning (ML), a subset of AI, teaching and learning environments.enables systems to analyze data, recognize patterns, and make II. REAL-WORLD CASE STUDIES ON AI-DRIVENdecisions with minimal human intervention. Deep Learning EDUCATION PLATFORMS(DL), a further subfield of ML, is inspired by the structure andfunction of the human brain and forms the backbone of many Several AI-driven education platforms have demonstratedAI applications, including generative AI. significant advancements in personalized learning and student

Collection

2025 Northeast Section Conference

Authors

Srilekha Bandla; Mukesh Reddy Jonnala; Peiqiao Wu; Sarosh Patel; Xingguo Xiong

D D = kd · W · Sf actor , Fpump = (17) must assess real-world performance, ensuring safety, re- T liability, and regulatory compliance.where kd = 0.1 mg/kg, W is patient weight, Sf actor represents According to the research findings, b etter s mart seizureseizure severity, and T is infusion time. This adaptive method detection intervention systems result from incorporating ma-prevents overmedication while ensuring controlled interven- chine learning with direct neurostimulation control and drugtion

Collection

2025 Northeast Section Conference

Authors

Dinh Cuong Nguyen; Dan Tenney

study EV adoption are diverse butArticles were critically evaluated for their application of underexplored. Many studies rely heavily on quantitativebehavioral theories, their methodological rigor in study designand analysis, and their practical relevance to real-world methods, with limited use of qualitative and mixed-methodchallenges. Specifically, the review assessed whether the studies approaches[8]. These gaps delay the development of effectiveeffectively utilized established behavioral theories (Theory of policies and strategies to promote EV adoptionPlanned Behavior, Technology Acceptance Model, and Theory of comprehensively.Reasoned Action) to

Collection

2025 Northeast Section Conference

Authors

Samuel Servati; PS. Dhanasekaran

, engineering concepts, preparing them for global industry and hands-on experimentation. Mathematics and demands. Evidence suggests that PBL enhances theoretical calculations remain essential, but by academic performance, retention of knowledge, and confidence in applying theoretical concepts to practical supplementing them with physical demonstrations and scenarios. Collaborative projects further develop real-world applications, students are better able to relate essential communication and teamwork skills while abstract equations to tangible engineering solutions [3]. promoting creativity and innovation through open ended This

Collection

2025 Northeast Section Conference

Authors

Lina H. Kloub, University of Connecticut; Christina Smith, University of Connecticut; Faiyhaa-Sydra Saulat, University of Connecticut

chine learning and data science principles. Participants engageassist in debugging code, optimizing designs, and automating in real-time applications of these methods, working on projectsdata analysis, making problem-solving more efficient. that require the modeling and simulation of complex systems AI presents dynamic, real-world problem scenarios that [3]. By utilizing this learning process, it leads to an enhance-significantly enhance engineering education. Adaptive coding ment of their ability to address intricate engineering problemsplatforms, for instance, utilize AI to create personalized learn- through AI-driven solutions, fostering engineer’s adept ating experiences by adjusting the complexity

Collection

2025 Northeast Section Conference

Authors

Shalini Jada; Xingguo Xiong; Ahmed El-Sayed; Navarun Gupta

- are displayed [2]. ment [23]. • Preprocessing Steps: Images are resized to 224x224 H. Validation and Testing pixels and normalized to ensure compatibility with the • Validation Dataset: A validation dataset is used to CNN input layer. Data augmentation techniques like evaluate the model’s real-world performance. Random flipping, rotation, and scaling are applied to improve images from the dataset are selected and predictions are model generalization [24]. compared to their ground truths [3].C. Model Development and Training • Performance Metrics: Accuracy, precision, recall

Collection

2025 Northeast Section Conference

Authors

Thomas C. McKinley

pursueprofessors who I was doing internship at Northeastern computer science in the future. The club has also fosteredUniversity as guest speakers. They provided insights into teamwork and problem-solving skills, which are valuablecurrent AI advancements and career opportunities in the fields outside of the field of programming.in real-world applications such as healthcare, finance, roboticsand natural language processing, which expanded our From a personal perspective, leading the club has providedunderstanding of the fields beyond simple coding exercises. valuable experience in organization, communication, andThis exposure inspired several students to pursue AI-related leadership. I have

Collection

2025 Northeast Section Conference

Authors

Marvin Gayle; Danny Mangra

. Problem-cellular, Bluetooth, wireless fidelity [Wi-Fi], or multiple other Based Learning (PBL) is a technique offered used to solvenetworks using various other protocols. An increased more technically complex challenges [5]. This approach isemphasis on real-world applications and collaboration with more focused and better suited for a collaborative or groupindustry partners allows a more tailored engineering learning approach, with opportunities to engage in critical thinking.experience [4]. Students can now interact with real-world lab There is a significant greater time investment required forequipment from remote location via cloud-based platforms PBL. This methodology can be less intensive at

Collection

2025 Northeast Section Conference

Authors

Lakshmi Aishwarya Malladi; Navarun Gupta; Ahmed El-Sayed; Xingguo Xiong

Mechanisms and VGG16." Neural Networks Journal, 2023.Nevertheless, issues including computing cost, datasetlimitations, and real-world applications must be resolved forgreater. .Future research should concentrate on enhancing the model foredge computing to allow real-time monitoring in resource-constrained contexts utilizing lightweight architectures such asMobileNet or Efficient Net. The integration of IoT sensors,drones, and cloud-based AI algorithms might establish anautomated and scalable early warning system. Furthermore,augmenting the dataset with real-time satellite imagery andsynthetic data would enhance model robustness andgeneralization across various terrains and

Collection

2025 Northeast Section Conference

Authors

Bushra Rayyan Qadri; Arthur McAdams; Peter Cavanaugh; Dan Tenney; Ali Baker

] called it The Battle for the Soul ofeach of these five courses, students in different American Business, where he referred to the financialdisciplines and backgrounds work together on team projects executives as “bean counters” and referred to the engineers asrelated to course material. These multidisciplinary teams “car guys” [4]. In this comparison, he makes the distinctionexperience real-world between those who value profits and cost cutting with those who prioritize design and innovation. The result was a big loss

Collection

2025 Northeast Section Conference

Authors

Peter F. Cavanaugh

classroom activities in academic courses [6- deeply embedded in industry practices, making them essential7]. for engineering and technology management students. Graduate-level education, particularly in Technology Both courses emphasize structured methodologies and data-Management, emphasizes the development of critical thinking driven decision-making. AI has the potential to transform theseand problem-solving skills. Courses like Project and Process areas by automating certain aspects of project planning, riskManagement also focus on the real-world application of these assessment, and process optimization. For example, AI-skills

Collection

2025 Northeast Section Conference

Authors

Teresa Piliouras; Steffi Crasto; Chinmay Dharap; Pui Lam Yu; Navarun Gupta

engagement is essential for sustaining interest evaluation and argument deconstruction; and fostering intellectual investment—particularly in o In workplace training, critical thinking programs subjects like critical thinking, digital literacy, and should be tailored to industry-specific needs to cybersecurity. Traditional lecture-based instruction ensure relevance and practical application. often fails to capture attention or demonstrate Healthcare training might emphasize diagnostic relevance. Educators can address this by emphasizing reasoning, data protection, and ethical decision- real-world applications, interactive problem-solving

Collection

2025 Northeast Section Conference

Authors

Ahmed Hassebo; Kevin B. Montes

connectivity to smart devices easily. Here, we employ thethe ESP32 to facilitate real-time data collection and control of ESP32-WROOM-32D, a generic high-performance Wi-Fi +irrigation processes. BT + BLE MCU module. It allows for a wide range of In this paper, the Smart Irrigation System operates in two applications from low power sensor networks up to real-timemodes: Manual Mode and Automatic Mode. Users can control data processing and thus is a general-purpose option for variousthe system using any device, whether it's an iPhone embedded systems. The ESP32-WROOM-32D module offers(iOS) or Android phone, through the Blynk application. The a 32 Mbit core running at

Collection

2025 Northeast Section Conference

Authors

Zebin Pepin; Douglas E. Dow

effective in small settings, it is labor-intensive and prone toindividuals in real-time, identifying discrepancies such as line- human error. The presence of staff may deter some individualscutting. Testing results demonstrated improved accuracy over from cutting in line, but it cannot entirely prevent ormanual queue monitoring, indicating potential for real-world accurately track such behavior, especially in large crowds [5].implementation. Further refinement is necessary to enhance For another method, some venues have adopted ticketingreliability across diverse operational environments. If fully systems that assign visitors to specific time slots. Thisdeveloped, such an automated queuing system could

Collection

2025 Northeast Section Conference

Authors

Yegin Genc; Gonca Altuger-Genc; Akin Tatoglu

level of AI obtain, especially for nonlinear systems or systems with complex boundaryinvolvement. Learning goals will allow for identifying the extentof the interaction (e.g., whether the topic needs to be broken intomore manageable learning units) and the learning structure (e.g., B. Assessing Learning Goalsmastering theoretical constructs vs. application of theory to real-world problems.) The AI tutor assists learners in defining their learning objectives, whether assigned by an instructor or self-decided. Goals may include mastering

Collection

2025 Northeast Section Conference

Authors

Joseph P. Duszak; John F. Drazan; Cynthia A. Bautista

burnout. Additionally, undergraduate nurses in training partaking in clinical rotations may observe practicing nurses using A common method to prepare future nurses for the rigors of strategies outside or against their academic training in high-the clinic is to provide real-life simulation laboratory (Sim- pressure situations, further widening the gap [5]. Sim-LabsLab) experiences where training institutions create simulated, have the potential to address the theory-practice gap byreal-world, healthcare spaces with accurate equipment, allowing nurses to apply their formal education to the clinicalsimulated patients, and

Collection

2025 Northeast Section Conference

Authors

Uma Balaji

script, providedThis is followed by practical measurements, where students through the learning management system, to visualize FMobserve the AM signal output from a function generator using signal plots with a sinusoidal tone as the modulating messagean oscilloscope and spectrum analyzer, reinforcing theoretical signal. They then solve an in-class problem on representingconcepts through real-world validation. Additionally, they the spectrum of modulated signals for a given modulationlearn to use MATLAB Simulink to generate DSB-SC index. This is followed by practical measurements using amodulated signals using Product, Sine Source, Oscilloscope, function generator, oscilloscope, and spectrum analyzer toand

Collection

2025 Northeast Section Conference

Authors

ZANNATUN NAYEEM; Tauhid Uddin Mahmood; Dan Tenney

. data analytics can achieve improved data processing,Furthermore, this research examines case studies of companies leading to better business insights and operationalthat have successfully addressed data integration challenges. efficiency.These case studies are actual applications that support the 5. Joel and Oguanobi (2024): Focuses on predictivevalue of best practices like real-time access to data, strong analytics for business expansion, demonstrating howdata validation procedures, and AI-powered analysis [9]. From data-driven strategies identify new marketthe compilation of these results, the research hopes to provide opportunities and enhance

Collection

2025 Northeast Section Conference

Authors

Shashi Kiran Chandrappa; Sidike Paheding; Yu Cai Cai

to improving web application security. By using inference onstudy employ a comparative analysis framework, using datasets datasets with vulnerability examples, these models can recog-of cookie security and cybersecurity logs, and apply prompt nize patterns and anomalies signaling potential security flaws.engineering to evaluate LLMs in identifying flaws in HTTP This method supports scalability and efficiency, allowing real-headers, analyzing security attributes. Our findings show LLMscan detect insecure cookie configurations, automate assessments, time monitoring and quick detection. Integrating LLMs intoand provide actionable insights, though challenges

Collection

2025 Northeast Section Conference

Authors

Rudra Mitra; Intiser Islam; Shariar Islam Saimon; Xingguo Xiong

multiple healthcare settings will contribute to the real- [3] Yoon, J., Jordon, J. and Schaar, M., 2018, July. Gain: Missing dataworld performance assessment. Real-world tests maintained by imputation using generative adversarial nets. In International conference on machine learning (pp. 5689-5698). PMLR.other datasets of patients will improve the trustworthiness of [4] Yan, K., Li, T., Marques, J.A.L., Gao, J. and Fong, S.J., 2023. A reviewclinical applications. on multimodal machine learning in medical diagnostics. Math. Biosci

Collection

2025 Northeast Section Conference

Authors

Bala Maheswaran; Meghna Sridhar; Yiannis Levendis; Hameed Metghalchi

mindset involves not only essential technologies. This interdisciplinary approachdesigning efficient systems but also considering their reflects the real-world complexity of energy systems, whereeffective solutions require collaboration across various Q1. Understanding Heliostat The heliostat model assembly helped me understand theareas of expertise. principles of concentrated solar power and its applications in renewable energy.Additionally, the heliostat model assembly

Collection

2025 Northeast Section Conference

Authors

Nusrat Zahan; Sidike Paheding

. Additionally, random saturation address this problem by gathering real-world datasets [6]–[8].contributed to the highest SSIM values, reaching 0.7456 with In many cases, acquiring such data on a large scale is oftenSRCNN on UFO-120 and 0.7570 with DEEP SESR on USR-248. Keywords—Image Super-Resolution, Data Augmentation, time-consuming and expensive. However, that is where DADeep Learning, PSNR, and SSIM. can play an important role, but only a few studies have been conducted [9], [10]. Radu et al. [10] investigated several meth- I. I NTRODUCTION

Collection

2025 Northeast Section Conference

Authors

Kalyan Khatry; Reihaneh Samsami

achievement [8]. C. Competency-Based Education Competency-based Education (CBE) focuses on demonstrable skills and real-world application rather than time- bound learning. Gen AI supports CBE by providing immediate, targeted feedback on practical tasks such as engineering design

Collection

2025 Northeast Section Conference

Authors

Mina Gaber Wahba Ibrahim; Xingguo Xiong; Navarun Gupta; Ahmed El-Sayed

Fig. 12. The figure shows the dry paper ignited readings in closetree with dry leaves, red cedar, and hemlock. These materials environment during combustion that are exponentially increased thewere chosen to simulate real world wildfire scenarios and sensor response.conditions, as wildfire often originates from trees and fromvegetation. The results of the dry paper are summarized. The high rise in particulate matter concentration and VOC levels demonstrated the system’s

Collection

2025 Northeast Section Conference

Authors

Shruti Brahma, University of New Haven; Siddhant Alhat Rajendra, University of New Haven; Ardiana Sula, University of New Haven

countries.ing data processing and analysis, and ARIMA for time-series Leveraging data from the Open Doors: International Stu-forecasting. But in order to provide a better performance on dents in the U.S. (1949–2023) report alongside admissionscomplex, non-linear relationships, we are moving towards using records from the University of New Haven, our model achievesRNN models. a prediction accuracy of 93%. To support these efforts, we de- The project’s interactive Power BI dashboard enables poli-cymakers and institutions to simulate real-world scenarios and veloped a data-driven decision

Collection

2025 Northeast Section Conference

Authors

Seyedeh Elham Kamali; Reihaneh Samsami

these skills by students. While the benefits of such technologies are providing simulation and modeling capabilities that numerous, they inevitably introduce complex ethical and allow students to engage with real-world scenarios in a integrity challenges that require constant supervision. controlled environment. This integration of AI can support a more dynamic and interactive learning 1experience that is aligned with industry demands (How engineering curricula at the University of New Haven.& Hung, 2019). The

Collection

2025 Northeast Section Conference

Authors

Paul Cotae; Nian Zhang; Onyinye Obioha-Val

screening are crucial to minimizing false positives and false results. Other real-world applications are in autonomous systems which include self-driving cars (for object detection) and Drones, which use the model for surveillance and threat detection. Banks and Insurance companies also used the model for fraud detection and insurance claims. All these are deployed

Collection

2025 Northeast Section Conference

Authors

Eric P. Flynn; Arthur McAdams

other industries (such as aerospace and biomedical) – to exploit A LLM is a type of AI designed to process and generate a AI LLM to assist a multidisciplinary team in the explorationhuman-like text by understanding and predicting patterns in of optimized material cost, engineering design, and ultimately,language. These models are built using large and diversified the real-world manufacturing of the part.amounts of text data and are trained on advanced machinelearning techniques, particularly deep learning algorithms, to A. From napkin sketch to CADunderstand context, syntax, semantics, and even nuances in One objective of AI is to assist users in completing taskslanguage. The scale of

Collection

2025 Northeast Section Conference

Authors

Hasan Baig; Phillip Bradford

, obtaining the correct answer; it's about understanding theacademicians are also readily using ChatGPT for different concepts, critical thinking, problem-solving, and developingpurposes. For example, in [13] Wang et.al., assessed skills that can be applied in real-world scenarios. OverrelianceChatGPT's ability to support various design, manufacturing, on AI models might hinder the development of these crucialand engineering education tasks. Their findings highlight skills.ChatGPT's impressive capacity to provide information, To address these concerns, educators are exploring variousgenerate structured content, and propose initial

Collection

2025 Northeast Section Conference

Authors

Boluwatife E. Faremi; Javier O. Pinzon-Arenas; Amir Mohammad Karimi Forood; Josef Kundrat; Hugo F. Posada-Quintero; Ann Marie Hoyt-Brennan; Wendy A. Henderson

platform's efficacy. Additionally, the artificialindicating increased cognitive effort due to the AIHT’s reduced laboratory setting may not fully capture the complexity of real-reliability. Simultaneously, performance accuracy of the world clinical decision-making environments that traineeparticipant dropped by 50%, suggesting that diminished AI nurses will encounter [1], [7]. Despite these limitations, thisaccuracy weakened trainee nurses’ trust, negatively affecting study represents an important step toward developing objectivetheir performance (Fig. 3b). measures of trust in healthcare AI. The AIHT platform's