Historical Data Research Articles

Formulating three classes of partial borrowing power priors to leverage historical data in process validation

The power prior approach has been widely used in decision-making processes across various fields where historical data is available. However, its application in process validation—a critical step in pharmaceutical manufacturing to ensure product quality and safety—remains limited, particularly in partial borrowing of historical information from early development batches. This manuscript addresses this gap by formulating and evaluating three classes of partial borrowing power priors: (PBPP1) a partial borrowing power prior with a fixed discounting parameter, (PBPP2) an unnormalized partial borrowing power prior with a random discounting parameter, and (PBPP3) a normalized partial borrowing power prior with a random discounting parameter. A Bayesian linear mixed model is employed for each class to account for both intra-batch and inter-batch variability in drug product potency. These methods are illustrated with simulated data based on real drug product data to provide a framework for leveraging historical data in process validation and to determine the number of samples per batch required to adequately characterize intra-batch variability for process performance qualification (PPQ). By modeling relevant data and conducting a simulation study, we confirm the limitations of PBPP1 and PBPP2 in addressing discrepancies between historical and current data, while demonstrating that PBPP3 effectively handles potential incompatibilities between historical and current data and makes inferences regarding the discounting parameter.

Reliability Analysis and Life Cycle Costing of Rooftop Solar Photovoltaic (PV) System Operating in a Composite Environment

Solar photovoltaic systems are becoming more and more common, so it's critical to understand how system or component failure impacts lifetime costs. Reliability analysis uses historical performance data to help identify equipment or systems that perform badly. A case study of solar PV systems' dependability and life cycle cost analysis is presented in this research. Manufacturers and consumers of solar PV systems provide the information needed for reliability study. ReliaSoft's Weibull++ software is used to estimate the best-fit failure distribution. Initial expenses, failure, operation and maintenance, and net salvage value are the categories into which the lifetime costs are divided. The findings of the analysis demonstrate that, from the standpoint of life cycle cost analysis, certain components like the inverter are essential. The Mean Time Between Failure (MTBF) of earthing and grounding, DC CB, IGBT, AC CB, Grid, AC converter, relay, inverter, cooling fan, and SMBUS communication components is lower; however, since they need only modest repairs, this expense is covered by routine inspection. Understanding how costs affect a product's whole life cycle is made easier with the use of the Life Cycle Cost (LCC) study. The cost of operation and maintenance (O&M) accounts for around 74% of the overall LCC. Thus, it may be inferred that the LCC of the solar PV system was driven by O&M and failure costs. It gives the user an idea that O&M expenses associated with panel cleaning can be substituted by manual cleaning for small solar PV systems like those examined in this paper. Since failure costs are lower than O&M costs, it is possible to lower this cost by scheduling an ideal technician visit for routine inspection.

AI-Powered Digital Twin Technology for Highway System Slope Stability Risk Monitoring

This research proposes an artificial intelligence (AI)-powered digital twin framework for highway slope stability risk monitoring and prediction. For highway slope stability, a digital twin replicates the geological and structural conditions of highway slopes while continuously integrating real-time monitoring data to refine and enhance slope modeling. The framework employs instance segmentation and a random forest model to identify embankments and slopes with high landslide susceptibility scores. Additionally, artificial neural network (ANN) models are trained on historical drilling data to predict 3D subsurface soil type point clouds and groundwater depth maps. The USCS soil classification-based machine learning model achieved an accuracy score of 0.8, calculated by dividing the number of correct soil class predictions by the total number of predictions. The groundwater depth regression model achieved an RMSE of 2.32. These predicted values are integrated as input parameters for seepage and slope stability analyses, ultimately calculating the factor of safety (FoS) under predicted rainfall infiltration scenarios. The proposed methodology automates the identification of embankments and slopes using sub-meter resolution Light Detection and Ranging (LiDAR)-derived digital elevation models (DEMs) and generates critical soil properties and pore water pressure data for slope stability analysis. This enables the provision of early warnings for potential slope failures, facilitating timely interventions and risk mitigation.

Retrospective Investigation of Pediatric Histoplasmosis Diagnosed at a Tertiary Children's Hospital in an Endemic Area Over 11 Years.

Most histoplasmosis data are from adults. We describe pediatric histoplasmosis diagnosed at a tertiary pediatric care center in an endemic area. We performed a retrospective chart review of patients birth-18 years seen in our institution (January 1, 2010 to August 15, 2022) with histoplasmosis identified by International Classification of Disease codes. We performed descriptive and univariate analyses of demographic, historical, clinical, laboratory, and treatment data. Of 122 patients, 26 had disseminated histoplasmosis (DH), 71 pulmonary/thoracic (P/T) disease, 21 had histoplasmosis as an incidental finding, and 4 were classified as other. Median age was 14 years (range 0.3-18.7 years); most were non-Hispanic White (70.5%) and city dwellers (77.1%). We noted neither seasonality nor disease category differences for age, race and ethnicity, residence, or environmental factors. Compared to P/T, DH more frequently had complex chronic conditions (73.1% vs 12.7%, P < .001), fever (88.5% vs 52.1%, P < .001), fatigue (76.9% vs 46.5%, P = .01), vomiting (53.8% vs 25.3%, P = .01), anemia (hemoglobin median 10.05 vs 12.5g/dL, P < .001), elevated sedimentation rate values (median 31 vs 29mm/h, P = .02), blood and urine Histoplasma antigen detection (63.6/84.6% vs 20.0/14.8%, P < .001), and antifungal treatment (100% vs 70.4%, P < .001). Patients with DH had longer antifungal treatment courses (399 vs 84 days, P < .001). Exposures were rarely recorded, but the most common was chickens. Chest radiographs were less likely to show adenopathy in DH compared to P/T (8.7% vs 55%, P < .001). Presentations, risk factors, diagnostics use/results, and therapy for pediatric histoplasmosis most often parallel those in adults.

Development of a reference database on historical body sizing and digital twins

PurposeHistorical body sizing systems are widely used in reconstructions to reproduce the shape of the human body. However, the systems are less informative than modern standards. The article aims to present a framework for processing historical sizing systems and to generalize 19th century adult female body sizes within one reference anthropometric database.Design/methodology/approachThe study adapted historical data on body sizing to the modern methodology of creating sizing systems. It systematized 248 typical sizes from 33 sources published in North America between 1875 and 1918. The common body measurements were examined by means of correlation analysis. The data was used to introduce new key dimensions and body sizes. Regression models for calculating secondary dimensions were established. A parametric modelling software program was used to generate avatars in accordance with the developed database.FindingsThe developed database presents 111 typical body sizes divided into five constitution and three height groups. The new database agrees with historical body sizes published in Europe. The differences between the developed database and modern standards were confirmed to be in line with changes in the shape of the body during the 19th and the 20th century.Originality/valueThe study provides historical reconstructions and exhibitions with a reliable and comprehensive source of anthropometric data, as well as a diverse range of virtual mannequins.

Innovative Accounting Practices: Bridging Traditional Methods with Modern Financial Reporting

In today's rapidly evolving financial landscape, the integration of traditional accounting methods with innovative practices has become imperative for organizations aiming to maintain relevance and achieve financial transparency. This study explores the transformative journey of accounting practices, emphasizing how innovation is reshaping the way financial information is recorded, processed, and reported. By bridging the gap between conventional approaches and modern financial reporting requirements, organizations can enhance decision-making, ensure regulatory compliance, and foster stakeholder confidence. Traditional accounting methods, characterized by manual processes and historical data orientation, have laid the foundation for financial integrity and reliability. However, the dynamic business environment necessitates the adoption of advanced tools and techniques such as automation, artificial intelligence, blockchain, and cloud-based systems. These innovations not only streamline operations but also enable real-time financial analysis, predictive insights, and error reduction, marking a significant departure from the limitations of traditional methods. The paper investigates key areas where innovation has disrupted conventional practices, including ledger management, auditing, taxation, and financial disclosures. It also addresses the challenges organizations face in transitioning to modern accounting systems, such as data security concerns, resistance to change, and the need for skilled professionals. Case studies from diverse industries are presented to highlight the tangible benefits of adopting innovative accounting practices. The findings underscore the importance of a hybrid approach, where traditional accounting principles are preserved for their reliability while being augmented with modern technologies to meet evolving business demands. This integration not only ensures accurate and efficient financial reporting but also empowers organizations to navigate complex financial ecosystems effectively. The study concludes by offering strategic recommendations for organizations to embrace innovative accounting practices while maintaining the core values of accuracy, transparency, and accountability. This research aims to provide valuable insights for academicians, practitioners, and policymakers in advancing the future of accounting.

Prognostic adjustment with efficient estimators to unbiasedly leverage historical data in randomized trials

Abstract Although randomized controlled trials (RCTs) are a cornerstone of comparative effectiveness, they typically have much smaller sample size than observational studies due to financial and ethical considerations. Therefore there is interest in using plentiful historical data (either observational data or prior trials) to reduce trial sizes. Previous estimators developed for this purpose rely on unrealistic assumptions, without which the added data can bias the treatment effect estimate. Recent work proposed an alternative method (prognostic covariate adjustment) that imposes no additional assumptions and increases efficiency in trial analyses. The idea is to use historical data to learn a prognostic model: a regression of the outcome onto the covariates. The predictions from this model, generated from the RCT subjects’ baseline variables, are then used as a covariate in a linear regression analysis of the trial data. In this work, we extend prognostic adjustment to trial analyses with nonparametric efficient estimators, which are more powerful than linear regression. We provide theory that explains why prognostic adjustment improves small-sample point estimation and inference without any possibility of bias. Simulations corroborate the theory: efficient estimators using prognostic adjustment compared to without provides greater power (i.e., smaller standard errors) when the trial is small. Population shifts between historical and trial data attenuate benefits but do not introduce bias. We showcase our estimator using clinical trial data provided by Novo Nordisk A/S that evaluates insulin therapy for individuals with type 2 diabetes.

Penerapan Artificial Neural Network untuk Memprediksi Persediaan Obat Esensial

The availability of essential medicines is a fundamental factor in ensuring high-quality healthcare services, especially in primary healthcare facilities such as Puskesmas. Inefficient drug inventory management can lead to various issues, including drug shortages that disrupt medical services and overstocking that may result in waste due to expiration. An accurate prediction system is essential to support more effective and efficient drug inventory planning. This study aims to analyze historical drug usage patterns to generate more accurate predictions. The research methodology includes problem identification, data collection, preprocessing, ANN architecture design, implementation, and system evaluation. Historical drug usage data from previous years is used for training and testing, with a division of 70% for training and 30% for testing. The backpropagation algorithm is applied to optimize the model by adjusting parameters such as the number of neurons in the hidden layer, learning rate, and activation function. The study results show that the ANN model with a 12-12-1 architecture achieves a high prediction accuracy, with a Mean Absolute Percentage Error (MAPE) of 2.13% for paracetamol stock. The developed MATLAB application provides an interactive platform for users to input historical data and obtain dynamic stock predictions. This system implementation is expected to help Puskesmas manage drug inventory more effectively, reduce the risks of shortages and overstocking, and improve efficiency in essential drug distribution. This study contributes to the field of health informatics by demonstrating the effectiveness of ANN in drug inventory prediction. Future research may explore hybrid machine learning models or integrate external factors, such as seasonal disease patterns and community demand levels, to enhance predictive accuracy and adaptability.

A Novel Transpiration Drought Index for Winter Wheat in the Huang-Huai-Hai Region, China: A Process-Based Framework Incorporating Improved Crop Water Supply–Demand Dynamics

Monitoring agricultural drought is crucial for mitigating yield losses in winter wheat, especially in the Huang-Huai-Hai (HHH) region of China. Current drought indices often fall short in accurately representing the water supply–demand dynamics for crops, neglect irrigation practices, and overemphasize drought intensity rather than its evolution and overall impact. To address these concerns, we developed a novel transpiration drought index utilizing the Water Balance for Winter Wheat (WBWW) model. This index integrated variations in atmospheric conditions, soil moisture conditions, crop resistance, and irrigation practices to enhance the evaluation of water supply and demand dynamics. The WBWW model was initially validated against field transpiration measurements, achieving an R2 of 0.7573, thereby confirming its reliability for subsequent analyses. To create a mechanistic understanding of crop water supply and demand, we adopted the reduction rate of actual and potential transpiration to identify drought events and constructed joint probability distributions of drought duration and severity using copulas. This led to the development of the Winter Wheat Drought Assessment Index (WDAI). The grade threshold for the WDAI was established based on historical drought data from the HHH region through a series of statistical threshold determination methods. Our findings showed that the WDAI successfully identified 87.36% of drought samples according to their recorded grades, with 97.13% within one grade of historical records. Comparative analyses with retained regional data and existing indices—the Crop Water Deficit Index (CWDI) and the Relative Soil Moisture Index (RSMI)—further demonstrated its effectiveness. Our study represents a robust tool for dynamic drought monitoring in the HHH region and offers critical insights into agricultural irrigation practices.

Factors associated with measles infection in Afghanistan surrounding the 2021 outbreak: A multivariable analysis

Abstract Background Despite development of a safe and effective vaccine, measles remains a risk in many low-income countries where vaccination is less available. Afghanistan’s recent measles epidemic remains understudied and data are needed to inform the public health response. This study investigated the demographic and epidemiologic characteristics associated with suspected and laboratory-confirmed measles infections surrounding the 2021 outbreak in Afghanistan. Methods We conducted a multivariable retrospective analysis of historical measles case data in Afghanistan from July 2019 to June 2023. Results Confirmed measles infections in Afghanistan increased by more than 200% since July 2021, and the majority of cases during the recent outbreak were laboratory-confirmed. Early infancy was a risk factor for confirmed measles during the recent outbreak, as compared to prior (p &amp;lt;0.001). Epidemiologic linkage was identified as protective (p &amp;lt;0.001), while sex (p = 0.796) and travel history (p = 0.615) where not predictive of risk. Conclusions Routinely collected measles surveillance data suggest that infancy and unknown travel status posed greater risk to confirmed measles infection in Afghanistan during the 2021 epidemic as compared to prior, and known exposure may lead to lower rates of formal laboratory diagnosis. Results also highlighted the geographic shift in cases before versus during the outbreak, as well as differences in distribution of suspected and laboratory confirmed cases throughout the epidemic.

Harnessing Machine Learning to Analyze Energy Generation and Capacity Trends in the USA: A Comprehensive Study

The constraints in conventional energy forecasting models in the USA are increasingly being appreciated as the energy landscape evolves. Such models are generally constructed on linear assumptions that do not capture energy generation and consumption patterns as dynamic and multi-dimensional. This research project aimed to develop effective machine-learning models to interpret historical trends in energy generation and capacity in the United States. With access to massive datasets for various energy sources—fuel-based to renewable and nuclear energy—this research endeavors to discover actionable information that can optimize energy output and improve grid stability. The dataset used is a comprehensive overview of energy production in America, integrating information from a variety of sources, ranging from fossil fuels to renewables and nuclear energy. This extensive dataset is sourced from reliable sources such as the U.S. Energy Information Administration (EIA) which supplies vast historical and current data on energy consumption and production patterns; the Federal Energy Regulatory Commission (FERC) which supplies information on regulatory frameworks and market forces that affect energy production; and smart grid management systems that deliver real-time data on energy flows and grid performance. Selecting appropriate machine learning models is crucial to process engineered features and derive actionable insights effectively. For this purpose, we use Logistic Regression, Random Forest Classifier, and the Support Vector Machines. The model performance table indicates how well three machine learning algorithms—Logistic Regression, Random Forest, and SVM—classify between renewable and non-renewable energy sources. All three models are highly accurate, with SVM having a slight edge, followed by Random Forest and lastly Logistic Regression. AI-powered energy forecasting is revolutionizing energy infrastructure planning in America through the provision of sophisticated information to guide investment in grid upgrades and expansion. As America transitions to dependence on renewable energy, AI-based forecasting is essential to optimize the integration of solar and wind energy production onto the grid. The variability of these renewable resources poses special challenges for grid managers, who need to balance demand and supply in real time. Moreover, AI can be used to identify gaps in renewable energy capacity and storage. The infusion of AI-based insights into the energy sector has far-reaching implications for U.S. policymakers, equipping them with fact-based tools to efficiently apply energy reforms. With evolving energy patterns, legislators need to adjust regulations to facilitate a seamless transition to a more sustainable energy system.

Pareto-optimal peer-to-peer risk sharing with robust distortion risk measures

Abstract We study Pareto optimality in a decentralized peer-to-peer risk-sharing market where agents’ preferences are represented by robust distortion risk measures that are not necessarily convex. We obtain a characterization of Pareto-optimal allocations of the aggregate risk in the market, and we show that the shape of the allocations depends primarily on each agent’s assessment of the tail of the aggregate risk. We quantify the latter via an index of probabilistic risk aversion, and we illustrate our results using concrete examples of popular families of distortion functions. As an application of our results, we revisit the market for flood risk insurance in the United States. We present the decentralized risk sharing arrangement as an alternative to the current centralized market structure, and we characterize the optimal allocations in a numerical study with historical flood data. We conclude with an in-depth discussion of the advantages and disadvantages of a decentralized insurance scheme in this setting.

Temporal representation learning enhanced dynamic adversarial graph convolutional network for traffic flow prediction

Accurate traffic flow prediction serves as the foundation for urban traffic guidance and control, playing a crucial role in intelligent transportation management and regulation. However, current methods fail to fully capture the complex patterns and periodic characteristics of traffic flow, leading to significant discrepancies between predicted and actual values. This gap hampers the achievement of high-precision forecasting. To address this issue, this paper proposes a temporal representation learning enhanced dynamic adversarial graph convolutional network (TRL-DAG). Our approach utilizes a temporal representation learning strategy with masked reconstruction for pre-training, aiming to extract temporal representations from contextual subsequences in historical traffic data. We further introduce a dynamic graph generation network to enhance the flexibility of graph convolution, enabling the model to capture dynamic spatiotemporal correlations by integrating both current and historical states. Additionally, we design an adversarial graph convolutional framework, which optimizes the loss through adversarial training, thereby reducing the trend discrepancy between predicted and actual values. Experiments conducted on six real-world datasets demonstrate that TRL-DAG outperforms existing state-of-the-art methods, achieving superior performance in traffic flow prediction.

Robust Dynamic Modeling of Bed Temperature in Utility Circulating Fluidized Bed Boilers Using a Hybrid CEEMDAN-NMI–iTransformer Framework

Circulating fluidized bed (CFB) boilers excel in low emissions and high efficiency, with bed temperature serving as a critical indicator of combustion stability, heat transfer efficiency, and pollutant reduction. This study proposes a novel framework for predicting bed temperature in CFB boilers under complex operating conditions. The framework begins by collecting historical operational data from a power plant Distributed Control System (DCS) database. Next, the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) algorithm is employed to decompose the raw signals into distinct modes. By analyzing the trade-offs of combining modes with different energy levels, data denoising and outlier reconstruction are achieved. Key features are then selected using Normalized Mutual Information (NMI), and the inflection point of NMI values is used to determine the number of variables included. Finally, an iTransformer-based model is developed to capture long-term dependencies in bed temperature dynamics. Results show that the CEEMDAN-NMI–iTransformer framework effectively adapts to diverse datasets and performs better in capturing spatiotemporal relationships and delivering superior single-step prediction accuracy, compared to Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and Transformer models. For multi-step predictions, the model achieves accurate forecasts within 6 min and maintains an R2 above 0.95 for 24 min predictions, demonstrating robust predictive performance and generalization.

Analyses of Variation Trends of Winter Cold Snaps in Subarctic and Arctic Alaska

Arctic Alaska is warming at twice the rate of the rest of the nation, severely impacting infrastructure built on permafrost. As winters warm, the effectiveness of thermosyphons used to stabilize foundations diminishes, increasing the risk of infrastructure failure. Because thermosyphons operate with the highest efficiency during winter cold snaps, studying the variation trends and patterns of winter cold snaps in Alaska is particularly important. To address this issue, this study analyzes the historical temperature data of four selected locations in Subarctic and Arctic Alaska, including Bethel, Fairbanks, Nome, and Utqiagvik. The winter cold snap is defined as a period when the average daily temperature drops below a specific site’s mean winter air temperature. The frequency, duration, and intensity of the winter cold snaps are computed to reveal their trends. The results indicate that the mean annual air temperature (MAAT) shows a warming trend, accompanied by sudden warming after 1975 for all study sites. The long-term average monthly air temperature also indicates that the most significant warming occurs in the winter months from December to March. While the frequencies of winter cold snaps remain relatively unchanged, the mean intensity and duration of cold snaps show a declining trend. Most importantly, the most intense cold snap during which the thermosyphons are the most effective is becoming much milder over time for all study sites. This study focuses specifically on the impact of changes in winter cold spells on thermosyphon effectiveness while acknowledging the complexity of other influencing factors, such as temperature differences, design features, coolant properties, and additional climatic parameters (e.g., wind speed, precipitation, and humidity). The data for this study were obtained from the NOAA NCEI website. The findings of this study can serve as a valuable reference for the retrofit or design of foundations and for decision making in selecting appropriate foundation stabilizing measures to ensure the long-term stability and resilience of infrastructure in permafrost regions. Moreover, the insights gained from this research on freeze–thaw dynamics, which are also relevant to black soils, align with the journal’s focus on sustainable soil utilization and infrastructure resilience.

An optimized informer model design for electric vehicle SOC prediction.

SOC prediction is of great value to electric vehicle status assessment. Informer model is better than other models in SOC prediction, but there is still a gap in practical application. Therefore, based on the health assessment algorithm, a new optimized Informer model is proposed to predict SOC. Firstly, the health assessment is carried out through the historical running data of the electric vehicle to obtain the health matrix. Then, the health matrix is used to improve Encoder and Decoder modules and improve the prediction accuracy and speed of Informer model. Subsequently, the health matrix is utilized to optimize the prediction logic, reduce the influence of truncation error, and further improve the SOC prediction accuracy. Finally, using the Informer model before and after optimization, SOC prediction is performed using four different datasets. The results indicate that after optimizing the En-De module of Informer, prediction accuracy improved by approximately 15%, with prediction speed increasing by about 100%. Furthermore, optimizing the prediction logic to reduce truncation error further enhanced Informer's prediction accuracy by around 20%.

Genomic Insights Into the Origin, Decline and Recovery of the Once Critically Endangered Iberian Lynx.

The Iberian lynx was at the brink of extinction by the year 2000 but has since then, and thanks to intensive conservation measures, gone through a remarkable recovery, providing a much-welcomed and encouraging conservation success story. Genetic issues have probably contributed to the decline in the past, and the genetic management of inbreeding and genetic diversity is likely contributing to its recent recovery. The species was an early adopter of genetic and genomic approaches, and the combination of an extreme decline, an intensive monitoring and management programme and extensive genomic resources and data makes the Iberian lynx an excellent model for conservation genomics. Here, we review how genetic and genomic data have contributed to the knowledge of the species evolutionary and demographic history, the evaluation of the genetic status of the species through time, including historical and ancient data, and how this information has prompted and guided conservation actions. In the process, genomics provided valuable insights into the dynamics of functional variation in bottlenecked populations and the consequences of intraspecific and interspecific admixtures. In more applied terms, the species is subjected to an ambitious genetic monitoring and management programme, covering captive, remnant and reintroduced populations, which has succeeded in improving the genetic status of the species and thereby contributed to its recovery. Current genomic work aims at expanding these contributions with novel genomic resources and data whilecapitalising on extensive demographic and genealogical data provided by the ongoing non-invasive genetic monitoring programme.

Advanced Applications of Python in Market Trend Analysis Research

Objective: This paper explores the advanced applications of Python in market trend analysis, combining time series analysis, machine learning, and deep learning techniques to construct an efficient market trend forecasting framework, thereby improving the scientific and accurate decision-making process. Methods: Empirical analysis is conducted using historical financial market data to construct three models: ARIMA, LSTM, and SVR. Python is used for data preprocessing, model training, and prediction evaluation. The models' accuracy, stability, and robustness are comprehensively compared using three evaluation metrics: Mean Squared Error (MSE), Mean Absolute Error (MAE), and Trend Consistency Rate (TCR). Results: The experiments demonstrate that the LSTM model performs best in terms of prediction accuracy, trend consistency, and robustness, with the lowest MSE (0.015), lowest MAE (0.088), and highest TCR (85.3%). The SVR model ranks second, while ARIMA performs weakly when dealing with nonlinear data. Conclusion: Python, with its powerful data analysis capabilities and ease of algorithm implementation, provides comprehensive support for market trend analysis. The LSTM model, due to its ability to model non-linear relationships and capture long-term dependencies, is suitable for complex trend forecasting and provides valuable insights for market decision-making.

Signal-Based Trading Strategy for SPY ETF: A Multiple Linear Regression Approach

This study develops a signal-based trading strategy for the SPDR S&amp;P 500 ETF Trust (SPY) using a multiple linear regression framework to analyze interrelationships between SPY and global equity indices across U.S., European, Asian, and Australian markets. By synthesizing historical pricing data from these major benchmarks, the model generates systematic trading signals through predicted price trajectories. In controlled training scenarios, the strategy achieved superior risk-adjusted returns compared to passive buy-and-hold approaches, demonstrating the value of cross-market signal integration. While the framework shows promise for algorithmic trading systems, the study acknowledges limitations in generalizing historical patterns to evolving market conditions. The findings highlight opportunities to enhance predictive accuracy through machine learning architectures capable of processing nonlinear market dynamics. These insights advance quantitative trading research by establishing methodologies for cross-market signal synthesis and proposing pathways to develop adaptive models for volatile capital markets.

AI-Based Credit Scoring Models in Microfinance: Improving Loan Accessibility, Risk Assessment, and Financial Inclusion

AI has made a huge impact in micro financing through the introduction of credit scoring models that have made loans easily accessible, reduced risks associated with lending while increasing the number and population covered by micro finance. The key traditional credit scoring measures that form the basis of conventional credit scoring models mainly base their decision on historical financial data, thus locking out people who do not have formal records with banks in the society. Symbolic models apply new sources of data, like use of mobile phones, transactions, social media accounts, and biometric data to make new credit scores for borrowers. The application of machine learning models will enhance lending decisions at the microfinance institutions, lower costs of operations, and reduce default risks. Nevertheless, there are certain drawbacks that are associated with the use of big data; issues such as data privacy, algorithms bias, meeting regulatory requirements, and technical constraints for MFIs. This paper aims at reviewing the advancement in credit scoring in the microfinance market, the techniques used to couple credit assessment using Artificial Intelligence, and the concerns towards the ethical practice of deploying Artificial Intelligence. It also provides examples of good practice in AI adoption and explores the potential of AI to increase financial inclusion. Despite the benefits that may be brought about by AI in assisting the process of micro financing, it is crucial to ensure fairness, transparency, and responsible use of the technology in microfinance.