Real-world Scenarios Research Articles

On the computation of robust examination timetables: methods and experimental results

AbstractWith ever-rising student numbers and an increasing shift towards more interdisciplinary study programs, the requirements for finding schedules for courses and exams become ever more complex. In real-world scenarios, the models used for calculating solutions to the course and the examination timetabling problem often must be provided to the students at the time of registration. In the field of curriculum-based course timetabling, timetables are calculated based on the structure of the study programs. For the examination timetabling problem, only a few papers focus on scheduling exams without registration data, as the requirements for exams are often more strict, or partial information is known from course registrations. In this paper we show that with the use of robustness techniques, we can also define the examination timetabling problem based on curricula. We introduce three robustness measures that address the inherent uncertainty when using the curriculum-based model. These robustness measures, along with other quality measures, are analyzed using a multi-objective simulated annealing algorithm. The results are compared on the Pareto front approximations found. We present a case study showing that, without a significant loss in solution quality, the chance is significantly reduced that rescheduling will be required after the exact numbers for the model are known.

Developing an efficient explainable artificial intelligence approach for accurate reverse osmosis desalination plant performance prediction: application of SHAP analysis

In recent decades, securing drinkable water sources has become a pressing concern for populations in various regions worldwide. Therefore, to address the growing need for potable water, contemporary water purification technologies can be employed to convert saline sources into drinkable supplies. Therefore, the prediction of important parameters of desalination plants is a key task for designing and implementing these facilities. In this regard, artificial intelligence techniques have proven to be powerful assets in this field. These methods offer an expedited and effective means of estimating effective parameters, thus catalyzing their implementation in real-world scenarios. In this study, the predictive accuracy of six different machine learning models, including Natural Gradient-based Boosting (NGBoost), Adaptive Boosting (AdaBoost), Categorical Boosting (CatBoost), Support vector regression (SVR), Gaussian Process Regression (GPR), and Extremely Randomized Tree (ERT) was evaluated for modelling the parameter of permeate flow as a key element in system efficiency, energy consumption, and water quality using six various input combinations of feed water salt concentration, condenser inlet temperature, feed flow rate, and evaporator inlet temperature. The next phase of this research employed the SHAP interpretability method to illustrate the impact of individual variables on the model's output. Moreover, the predictive performance of the developed frameworks was evaluated using a set of five dependable statistical measures: RMSE, NS, MAE, MAPE and R 2. These indicators were utilized to provide a robust means of gauging the precision of the model's forecasts. A comparative analysis of the outcomes, as measured by the RMSE criteria, revealed that the SVR technique (RMSE = 0.125 L/(h·m2)) exhibited superior performance compared to NGBoost (RMSE = 0.163 L/(h·m2)), AdaBoost (RMSE = 0.219 L/(h·m2)), CatBoost (RMSE = 0.149 L/(h·m2)), GPR (RMSE = 0.156 L/(h·m2)), and ERT (RMSE = 0.167 L/(h·m2)) methodologies in predicting permeate flow rates. The modelling outcomes obtained during the evaluation stage demonstrated the efficacy of the SVR algorithm in enhancing the precision of permeate flow forecasts, utilizing relevant input variables.

A portfolio optimization model under uncertain random environment

<p>Uncertain events frequently occur in today’s financial markets. Consequently, the issue of portfolio selection is becoming increasingly significant. This paper thoroughly considers the complexities of stock returns in real-world scenarios and employs uncertain differential equations (UDE), uncertain time series analysis (UTSA), stochastic differential equations (SDE), and random time series analysis (RTSA) to predict stock returns, thereby enhancing the accuracy of these predictions. Furthermore, this paper addresses investors’ preferences and the limitations of using variance as a measure of investment risk. It introduces a risk preference factor and proposes an uncertain random mean-lower variance model. Finally, a genetic algorithm is utilized to solve the model, and numerical simulations are conducted to demonstrate the model’s practicality.<strong></strong></p>

Two-dimensional network scene design for traffic simulation: An approach using computational Delaunay triangulation

Traffic network design is a pivotal aspect of urban planning, necessitating a harmonious blend of theoretical and practical approaches. In this paper, we propose a novel network design methodology utilizing Delaunay triangulation. We address the limitations of raw Delaunay networks, particularly their triangular configurations, by introducing edge-reduction techniques guided by a newly developed metric, detour redundancy. This metric is instrumental in assessing edge importance, facilitating the transformation of the network to better mirror real-world traffic scenarios. The efficacy and practical application of the designed networks are then evaluated through traffic simulations using the principles of cellular automata. Our findings underscore the potential of this approach in enriching traffic network design and pave the way for future investigations into diverse network characteristics and their implications in urban environments.

Update of prognosis and characteristics of chronic obstructive pulmonary disease in a real-world setting: a 5-year follow-up analysis of a multi-institutional registry.

We conducted a prospective observational study to elucidate the long-term prognosis and management of chronic obstructive pulmonary disease (COPD) in clinical practice in Japan in the mid-2010s. This prospective cohort study included 29 facilities. Data from 427 patients clinically diagnosed with COPD, enrolled between September 2013 and April 2016, were analyzed. Interstitial pneumonia was excluded through a central multidisciplinary discussion. Follow-up data were collected for up to 5 years after patient registration. At the time of registration, 53 patients clinically diagnosed with COPD did not have airflow limitation (AFL). In the cohort with AFL (n = 374), 232 patients completed a 5-year follow-up, while 49 patients died during the 1576.6 person-years of observation. The mean age was 71.7 years with an overall 5-year survival rate of 85.4%. Stratified by % forced expiratory volume in one second (FEV1), survival rates were 93.6% in the mild and moderate AFL group, 82.5% in the severe AFL group, and 66.1% in the very severe AFL group. The prognosis of the subpopulation without AFL was poor with a 5-year survival of 81.6%. This subpopulation exhibited respiratory symptoms, low vital capacity and total lung capacity, and emphysematous changes. Our study presents the 5-year survival and real-world clinical practice scenario of a prospective cohort of patients clinically diagnosed with COPD in Japan in the mid-2010s. The survival rates of our cohort were numerically better than the Japanese cohort in the 1990s, regardless of the high median age of this cohort. Overall, 12.4% of the patients in this cohort with no AFL at registration exhibited respiratory symptoms and distinct spirometric patterns, and had a poor prognosis.

Cost-effectiveness of radiofrequency echographic multi-spectrometry (REMS) for the diagnosis of osteoporosis in the United States

Abstract Radiofrequency echographic multi spectrometry (REMS) is an innovative, non-ionizing diagnostic technique that has shown high accuracy and precision, making it a promising alternative to dual-energy X-ray absorptiometry (DXA) for osteoporosis diagnosis in clinical settings. With economic considerations playing an increasingly crucial role in healthcare decisions, this study aims to evaluate the cost-effectiveness and economic impact of improved osteoporosis diagnosis using REMS followed by treatment in the United States (US). A microsimulation-based Markov model was constructed to estimate the cost per quality-adjusted life year (QALY) gained (in US$2022) for REMS followed by treatment versus no diagnosis and treatment in US women aged 50 years and older with osteoporosis. Women were categorized as high risk (receiving alendronate monotherapy for 5 years) or very high risk (receiving an 18-month course of anabolic treatment, abaloparatide, followed by 5 years of alendronate). The study evaluated two medication adherence scenarios: one assuming full adherence to treatment and the other reflecting real-world adherence. The results indicate that REMS followed by treatment is associated with improved health outcomes, including more QALYs and fewer fractures, and reduced fracture-related costs compared to no diagnosis and treatment. The incremental cost-effectiveness ratio of REMS was estimated at $33 891 and $49 198 per QALY gained, under the full adherence and real-world adherence scenarios, respectively. These values are below the US cost-effectiveness thresholds of $100 000 to $150 000 per QALY. Moreover, a 5% increase in the diagnosis and treatment of women over 50 years at high and very high risk of fractures using REMS is projected to save approximately 30 000 life years, 43 500 QALYs, and prevent 100 000 fractures over a lifetime under real-world medication adherence. In conclusion, this study suggests that REMS is a cost-effective strategy for the diagnosis and management of osteoporosis in US women, offering substantial potential economic benefits and improved health outcomes.

Rolling Bearing Fault Diagnosis Model Based on External Attention Integrated Convolutional Neural Network under Imbalanced Data Conditions

Abstract Rolling bearings are essential components in numerous mechanical systems, and their failure can result in considerable downtime and expensive repairs. Therefore, accurate and timely fault diagnosis is vital for effective predictive maintenance and overall reliability. Traditional diagnostic methods often struggle with complex and non-stationary signals, compounded by issues of data imbalance in 
 realworld scenarios. A method for diagnosing rolling bearing faults has been developed in this paper utilizing External Attention (EA), Convolutional Neural Networks (CNN), and Continuous Wavelet Transform (CWT), specifically addressing the challenge of imbalanced sample data. This approach offers significant advantages, including a reduction in complexity by eliminating the need for data augmentation and leveraging external attention for enhanced feature extraction from samples. Compared to other attention mechanisms, this method demonstrates outstanding performance on both training and testing sets with imbalanced samples, exhibiting minimal overfitting tendencies. The proposed CWT-EACNN method effectively addresses the challenge of imbalanced sample data in rolling bearing fault diagnosis, demonstrating exceptional performance and reduced complexity.

Leveraging Mixture of Experts and Deep Learning-Based Data Rebalancing to Improve Credit Fraud Detection

Credit card fraud detection is a critical challenge in the financial sector due to the rapidly evolving tactics of fraudsters and the significant class imbalance betweenegitimate and fraudulent transactions. Traditional models, while effective to some extent, often suffer from high false positive rates and fail to generalize well to emerging fraud patterns. In this paper, we propose a novel approach that integrates a Mixture of Experts (MoE) model with a Deep Neural Network-based Synthetic Minority Over-sampling Technique (DNN-SMOTE) to enhance fraud detection performance. The MoE modeleverages multiple specialized expert networks, each trained to detect specific types of fraud, while the DNN-SMOTE generates high-quality synthetic samples to address the class imbalance. Our experimental results on a publicly available dataset demonstrate that the proposed method achieves a classification accuracy of 99.93%, a true positive rate of 84.69%, and a true negative rate of 99.95%. The Matthews Correlation Coefficient (MCC) of 0.7883 further highlights the model’s balanced performance in detecting fraudulent transactions. These results underscore the effectiveness of combining MoE with DNN-SMOTE, offering a robust solution for real-world credit card fraud detection scenarios.

Harnessing Deep Learning for Enhanced MPPT in Solar PV Systems: An LSTM Approach Using Real-World Data

Maximum Power Point Tracking (MPPT) is essential for maximizing the efficiency of solar photovoltaic (PV) systems. While numerous MPPT methods exist, practical implementations often lean towards conventional techniques due to their simplicity. However, these traditional methods can struggle with rapid fluctuations in solar irradiance and temperature. This paper introduces a novel deep learning-based MPPT algorithm that leverages a Long Short-Term Memory (LSTM) deep neural network (DNN) to effectively track maximum power from solar PV panels, utilizing real-world data. The simulations of three algorithms—Perturb and Observe (P&O), Artificial Neural Network (ANN), and the proposed LSTM-based MPPT—were conducted using MATLAB (2021b) and RT_LAB (24.3.3) with an OPAL-RT simulator for real-time analysis. The data used for this study were sourced from NASA/POWER’s Native Resolution Daily Data of solar irradiation and temperature specific to Imphal, Manipur, India. The obtained results demonstrate that the LSTM-based MPPT system achieves a superior power tracking accuracy under changing solar conditions, producing an average output of 74 W. In comparison, the ANN and P&O methods yield average outputs of 57 W and 62 W, respectively. This significant improvement, i.e., 20–30%, underscores the effectiveness of the LSTM technique in enhancing the power output of solar PV systems. By incorporating real-world data, valuable insights into solar power generation specific to the selected location are provided. Furthermore, the outputs of the model were verified through real-time simulations using the OPAL-RT simulator OP4510, showcasing the practical applicability of this approach in real-world scenarios.

Computational Math Modeling and AI Optimization for Better Decision-Making: Applications in Machine Learning

The integration of computational mathematical modeling, optimization, machine learning (ML), data analysis, and artificial intelligence (AI) offers transformative opportunities to solve complex problems across diverse fields. This paper delves into the development and application of advanced modeling techniques that blend traditional mathematical approaches with advanced AI algorithms to enhance predictive accuracy, optimize resource allocation, and improve decision-making processes in areas such as logistics, finance, engineering, and healthcare. By leveraging data-driven insights, these models can simulate real-world scenarios and identify optimal solutions more effectively than conventional methods. The paper also explores how AI-enhanced modeling can impact broader systems by reshaping industry practices, influencing frameworks, and potentially challenging established societal norms. Eventually, the paper argues for a balanced and informed deployment of AI-driven modeling techniques to maximize their benefits while addressing potential risks and limitations.

Artificial Intelligence as Research Methods in Urban Design

This paper delves into the role of artificial intelligence (AI) in urban design, focusing on its capability to manage complex urban systems via paradigm classification and comparative analysis. It methodically explores how AI improves design generation, optimizes morphology, and simulates real-world scenarios, thereby enhancing design quality and efficiency throughout the urban design process. Furthermore, the study contributes to the development of comprehensive AI evaluation criteria, integrating both theoretical and practical perspectives to critically understand the advantages and limitations of AI in urban design applications.

Corporate Social Responsibility and Virtual Reality: A Synergistic Approach to Building Sustainable Brand Image

This study explores the intersection of Corporate Social Responsibility (CSR) and Virtual Reality (VR) as a novel approach to fostering sustainable brand image. Utilizing a qualitative research method through literature review and library research, the paper analyzes existing frameworks and theories on CSR and their potential enhancement through immersive VR experiences. The findings indicate that CSR initiatives, when integrated with VR, offer companies a unique opportunity to engage consumers more deeply by creating interactive and impactful experiences that highlight their commitment to sustainability. VR can simulate real-world scenarios of a company’s social and environmental efforts, making these initiatives more tangible and relatable to consumers. This immersive engagement not only enhances consumer understanding of CSR efforts but also builds stronger emotional connections, which are essential for long-term brand loyalty. Additionally, the synergy between CSR and VR contributes to a more transparent and authentic brand image, as VR allows for a more compelling demonstration of CSR activities. The study concludes that this innovative approach of combining CSR with VR can significantly improve brand perception, foster consumer trust, and establish a sustainable brand image that aligns with modern consumer expectations of corporate responsibility and ethical behavior. Further research is suggested to explore the long-term effects of this integration on consumer behavior and brand equity

EEG Data Augmentation Method for Identity Recognition Based on Spatial–Temporal Generating Adversarial Network

Traditional identity recognition methods are facing significant security challenges due to their vulnerability to leakage and forgery. Brainprint recognition, a novel biometric identification technology leveraging EEG signals, has emerged as a promising alternative owing to its advantages such as resistance to coercion, non-forgeability, and revocability. Nevertheless, the scarcity of high-quality electroencephalogram (EEG) data limits the performance of brainprint recognition systems, necessitating the use of shallow models that may not perform optimally in real-world scenarios. Data augmentation has been demonstrated as an effective solution to address this issue. However, EEG data encompass diverse features, including temporal, frequency, and spatial components, posing a crucial challenge in preserving these features during augmentation. This paper proposes an end-to-end EEG data augmentation method based on a spatial–temporal generative adversarial network (STGAN) framework. Within the discriminator, a temporal feature encoder and a spatial feature encoder were parallelly devised. These encoders effectively captured global dependencies across channels and time of EEG data, respectively, leveraging a self-attention mechanism. This approach enhances the data generation capabilities of the GAN, thereby improving the quality and diversity of the augmented EEG data. The identity recognition experiments were conducted on the BCI-IV2A dataset, and Fréchet inception distance (FID) was employed to evaluate data quality. The proposed method was validated across three deep learning models: EEGNET, ShallowConvNet, and DeepConvNet. Experimental results indicated that data generated by STGAN outperform DCGAN and RGAN in terms of data quality, and the identity recognition accuracies on the three networks were improved by 2.49%, 2.59% and 1.14%, respectively.

Comparative Sensitivity Analysis in Composite Material Selection: Evaluating OAT and COMSAM Methods in Multi-criteria Decision-making

The decision-making process is critical across various fields, influencing essential aspects like performance, safety, and sustainability. In engineering and materials science, selecting suitable materials is a complex process involving multiple interdependent criteria, which impact the reliability and lifecycle of a design. Traditional sensitivity analysis methods, such as the one-at-a-time (OAT) approach, assess the effects of individual parameter changes but often fail to capture the cumulative impact of simultaneous modifications. It can limit the effectiveness of decision tools in real-world scenarios where multiple factors vary concurrently. This study utilizes the Comprehensive Sensitivity Analysis Method (COMSAM) to address limitations in traditional sensitivity analysis by modeling simultaneous parameter adjustments across multiple criteria. Specifically, the study aims to validate COMSAM’s effectiveness in the practical problem of composite material selection, comparing it with the traditional OAT approach within a multi-criteria decision-making (MCDM) framework. By integrating both sensitivity analysis techniques with the Characteristic Objects Method (COMET), this research ensures robust evaluation and mitigates vulnerabilities like the rank reversal paradox. The findings demonstrate that COMSAM provides enhanced insights into the impact of parameter interdependencies, offering a more resilient foundation for decision-making in high-stakes environments where material properties and performance are paramount.

Intrusion Detection System in Network Security Using Naive Bayes and Support Vector Machine

An Intrusion Detection System (IDS) is designed to detect suspicious activities or security threats within a network, necessitating continuous advancements in the field. Both implementation techniques and algorithmic research play pivotal roles in enhancing IDS capabilities. This study addresses this need by focusing on the implementation and comparison of two prominent classification models: Naive Bayes and Support Vector Machine (SVM). The study is centered within the domain of Intrusion Detection System (IDS) tailored for network security. In the course of this research, a relevant dataset sourced from Kaggle serves as the foundation for training and testing both classification models. The findings of this study underscore the models' efficacy in intrusion detection. The SVM model, in particular, emerges as a standout performer, showcasing an accuracy rate that approaches 100%, thus exemplifying its potential in real-world scenarios. Meanwhile, the Naive Bayes model delivers commendable accuracy, surpassing 88%. This investigation not only contributes to the advancement of intrusion detection methodologies but also highlights the viability of these classification models for bolstering network security against the ever-evolving threat landscape.

Unveiling trends in migration of iron-based nanoparticles in saturated porous media

Nanoscale zero-valent iron (nZVI) particles are routinely used for environmental remediation, but their transport dynamics in different settings remain unclear, hindering optimization. This study introduces a novel approach to predicting nZVI transport in saturated porous model environment. The method employs advanced long column devices for real-time monitoring via controlled magnetic susceptibility measurements. Numerical modeling with a modified version of the MNMs 2023 software was then used to predict nZVI and its derivatives mobility in field-like conditions, offering insights into the radius of influence (ROI) and shape factor (SF) of their distribution. A standard nZVI precursor was compared with its four major commercial derivatives: nitrided, polyacrylic acid-coated, oxide-passivated, and sulfidated nZVI. All these iron-based nanoparticles exhibited identical particle sizes, morphologies, surface areas, and phase compositions, isolating surface properties, dominated by charge, as the sole variable affecting their mobility. The study revealed optimal transport when the surface charge of nZVI and its derivatives was strongly negative, while rapid aggregation of nZVI derivatives due magnetic attraction reduced their mobility. Modeling predictions based on column scale-up, indicated that detectable concentrations of 20 g L⁻1 were found at distances ranging from 0.4 to 1.1 m from the injection well. Slightly sulfidated nZVI traveled farther than the nZVI precursor and ensured more homogenous particle distribution around the well. Organically modified nZVI migrated the longest distances but showed particle accumulation close to the injection point. The findings suggest that minimal sulfidation combined with organic modification of nZVI surfaces may effectively enhance radial and vertical nZVI distribution in aquifers. Such improvements increase the commercial viability of modified nZVI, reduce their adverse impacts, and boosts their practical applications in real-world scenarios.

Electroencephalogram Emotion Recognition via AUC Maximization

Imbalanced datasets pose significant challenges in areas including neuroscience, cognitive science, and medical diagnostics, where accurately detecting minority classes is essential for robust model performance. This study addressed the issue of class imbalance, using the ‘liking’ label in the DEAP dataset as an example. Such imbalances were often overlooked by prior research, which typically focused on the more balanced arousal and valence labels and predominantly used accuracy metrics to measure model performance. To tackle this issue, we adopted numerical optimization techniques aimed at maximizing the area under the curve (AUC), thus enhancing the detection of underrepresented classes. Our approach, which began with a linear classifier, was compared against traditional linear classifiers, including logistic regression and support vector machines (SVMs). Our method significantly outperformed these models, increasing recall from 41.6% to 79.7% and improving the F1-score from 0.506 to 0.632. These results underscore the effectiveness of AUC maximization methods in neuroscience research by offering a robust solution for managing imbalanced datasets, developing more precise diagnostic tools and interventions for detecting critical minority classes in real-world scenarios.

Study of the Framework and Attributes of Ethical Standards in Journalism

It has always been difficult to create fair evaluation criteria for journalists, especially when it comes to judging morality and ethical conduct in news reporting. These guidelines function as yardsticks for gauging the ethics of news topics and the objectivity of journalistic techniques. These criteria are becoming more and more difficult as journalism changes, especially in the digital age. Fostering ethical journalism requires an understanding of its categorization and internal structure. This analysis employs a conceptual framework to explore the evaluation standards of news ethics. The method involves: Literature Review: Examining existing academic literature, ethical codes, and industry guidelines to outline the prevailing evaluation standards in journalism. Thematic Analysis: Identifying and categorizing the components of evaluation standards into normative and quality standards, assessing their relevance to different aspects of journalistic practice. Case Studies: Analyzing specific instances where journalistic ethics were put to the test, allowing for a practical understanding of how these standards function in real-world scenarios. The primary objective of this study is to clarify the internal structure of journalistic ethics evaluation standards and highlight the distinctions between normative and quality standards. By doing so, the analysis aims to provide insights into how these standards can be effectively applied to evaluate news behaviors and moral character in journalism. What is deemed ethical in one context may not hold the same weight in another. This relativity underscores the importance of understanding the specific circumstances surrounding journalistic practices. Historical Context: The evolution of journalistic ethics is tied to historical events and societal changes. Standards that once sufficed may need reevaluation to address contemporary challenges, particularly with the rise of digital media and misinformation. Complex Internal Structure: The internal structure of evaluation standards can be multi-layered, with different levels of criteria applied depending on the context and specific ethical dilemmas faced by journalists. This complexity requires continuous dialogue and adjustment as new ethical challenges arise. Practical Implications: Journalists and media organizations must engage with these standards actively, incorporating them into training and policy development. By fostering an understanding of both normative and quality standards, the industry can promote a more ethical and accountable journalism practice.

The Use of Generative Artificial Intelligence for Business Decision-Making

The purpose of this study was to develop autonomous artificial intelligence agents capable of working as a cohesive group to solve intricate decision-making processes, ultimately pushing the boundaries of artificial intelligence’s applicability in high-stakes and real-world scenarios. An autonomous multi-agent model was designed and deployed using the Analytic Hierarchy Process (AHP) decision-making framework, with autonomous AI agents, such as CrewAI and AutoGen, orchestrating multiple stages of activities performed by four distinct agents leveraging generative AI (GenAI). These agents were applied to a use case for proof of concept. The agents used a qualitative dataset to generate management recommendations based on cost and market feasibility considerations. The model was tested with publicly available customer feedback data on oatmeal cookies. It synthesized 913 customer reviews, identified common complaints, provided a summary of potential solutions, and generated a summary of market opportunities, along with relevant challenges. Although a subject matter expert's review is necessary to evaluate the practicality and relevance of the recommendations, the results demonstrated the high potential of multi-agent models in synthesizing and distilling large datasets into actionable insights, thereby augmenting decision-making processes in business contexts. The model's modular design allows for enhancements in quality, accuracy, and practicality by incorporating additional datasets and new agents. This study underscores that the integration of GenAI within a multi-agent model empowers businesses to swiftly transform vast amounts of business intelligence data into practical recommendations.

Constructing a multifunctional reactive MXene-based additive for micro-crosslinking polyurea elastomer to achieve superior mechanical properties and enhanced fire safety

The widespread application of polyurea (PUA) in protective coatings typically requires excellent flame retardancy due to real-world usage scenarios. However, the addition of flame retardant additives often compromises the material’s mechanical properties. In this study, a multifunctional reactive additive, MX-SP-NH2, was creatively prepared by covalently grafting spirocyclic pentaerythritol bisphosphorate disphosphoryl chloride (SPDPC) onto MXene (Ti3C2Tx) surfaces, followed by the introduction of 1,3-propanediamine (PDA) by the reaction with SPDPC. MX-SP-NH2 was then incorporated into PUA, participating in its crosslinking process and significantly enhancing both its mechanical strength and flame retardancy. Compared to the control PUA sample, the micro-crosslinked PUA sample with just 1.0 wt% MX-SP-NH2 demonstrated a 184.3% increase in tensile strength, a 126.1% increase in elongation at break, and a 436.3% increase in toughness. The flammability characterization revealed that the peak heat release rate, peak smoke production rate, and peak CO production rate were reduced by 31.7%, 28.9%, and 61.5%, respectively, indicating a substantial improvement in fire safety. Additionally, the flame retardant PUA-based flexible strain sensors demonstrated stable electrical signal responsiveness. This work has opened up new potential applications of PUA in flexible electronics.