This data paper documents and evaluates five publicly accessible workplace injury datasets: Occupational Safety and Health Administration (OSHA) Severe Injury Reports (SIR), OSHA Injury Tracking Application (ITA) Case Detail, Texas Institutional Medical Billing Services (SV2), Oregon Workers’ Compensation Record Level Claims (OR WC), and New York State Workers’ Compensation Injury Claims (NY WC). Because occupational injury data are fragmented across jurisdictions and sources, analyzing dataset structure and harmonizing key fields is necessary to enable consistent cross-dataset analytics and modeling. These data sources have a combined total of over 7.5 million records from 2015 to 2024 across federal and state jurisdictions. The primary focus of this study is to introduce and provide detailed dataset analyses, including data structures, reporting mechanisms, features, and exploratory patterns useful for occupational safety research and modeling. To follow, we present a cross-dataset harmonization (schema alignment) process that integrates four incident- and claim-level datasets (SIR, ITA, OR WC, NY WC) using a constrained set of shared features. To demonstrate the utility of the harmonized dataset, we include a baseline decision tree classifier as a machine learning demonstration, showing generalizable predictions for fall-related injuries across jurisdictions. Overall, our work provides a detailed reference for researchers seeking to understand, compare, and work with heterogeneous workplace injury data for analytics and machine learning.

This survey paper explores the application of machine learning (ML) techniques in the domain of workplace injuries, focusing on three key areas: risk prediction, return to work (RTW) strategies, and demographic analysis. Through an extensive review of literature from January 2015 to July 2024, this paper examines the latest advancements in ML-driven approaches to workplace safety and identifies important research gaps. This paper highlights how classical ML techniques, such as ensemble models and decision trees, have become essential tools for identifying workplace injury risks, enabling more accurate interventions. It emphasizes the importance of leveraging ML in personalized RTW programs, which use data-driven insights to improve recovery outcomes and reduce economic demands. In the context of demographic analysis, this paper explores how ML algorithms can uncover disparities in injury rates across various age groups, industries, and occupations, underscoring the need for targeted safety measures. Moreover, research gaps are identified, particularly regarding the emerging potential of advanced ML techniques, such as deep learning and large language models (LLMs), for analyzing structured and unstructured safety data, methods that have not yet been widely applied in workplace injury research. As such, future research should apply recent advances in ML, integrating these approaches with comprehensive and accessible datasets to enhance the prediction and prevention of workplace injuries, provide more detailed analytics and insights, and improve safety protocols across all industries. This comprehensive survey is an invaluable resource for researchers and practitioners leveraging ML to address complex challenges in workplace safety.

Medicare fraud in the U.S. is a widespread issue that undermines the integrity of the Medicare system, leading to wasted resources and significant financial losses. To combat Medicare fraud, we apply advanced machine learning techniques to improve detection accuracy and effectiveness. Our study introduces a feature selection framework that incorporates SHapley Additive exPlanations (SHAP) with algorithms such as Light-GBM, XGBoost, CatBoost, and Decision Tree as base learners to calculate SHAP values for selecting the best feature subsets. Our experiments utilize the Medicare Part D and Part B datasets accessible from the Centers for Medicare and Medicaid Services (CMS) website. These datasets are known for issues related to class imbalance and high dimensionality, which pose challenges for building classification models. To address these challenges, we implement Random Undersampling (RUS), followed by SHAP to select the most informative features. These data subsets are then evaluated using a Random Forest classifier, with the Area Under the Precision-Recall Curve (AUPRC) serving as an evaluation metric for model performance. Our results reveal that combining RUS with the SHAP-based feature selection framework generates feature subsets that yield models with statistically similar or improved performance compared to those without the RUS enhancement. Furthermore, by reducing the need for computational resources, our approach improves efficiency while maintaining model performance at similar or enhanced levels.

Fraud detection through the classification of highly imbalanced Big Data is an exciting area of Machine Learning research. On the one hand, in certain fraud detection application domains, the use of One-Class classifiers is an overlooked opportunity. On the other hand, for researchers faced with the task of building Machine Learning models for identifying fraud, when only legitimate transaction data is available, One-Class Classifiers are indispensable. We investigate the efficacy of SHapley Additive exPlanations (SHAP) as a feature selection technique for One-Class classification tasks. In this study we utilize authentic data from the Credit Card fraud and Medicare insurance fraud application domains. Our contribution is to show that researchers can use SHAP in conjunction with One-Class Classifiers to do feature selection on highly imbalanced datasets, and then build models, with the selected features, that yield performance similar to, or better than, models built using all features. Our results in Big Medicare data fraud detection show that an over 90% data reduction through feature selection can nevertheless coincide with the best performance in terms of Area under the Precision Recall Curve.

SHapley Additive exPlanations (SHAP) is a method used to explain the output of machine learning models. SHAP provides a unified measure of feature importance through SHAP values and serves as a feature selection tool for handling Big Data. This paper presents a study on optimizing feature selection using SHAP for Medicare fraud detection by applying the Random Undersampling (RUS) technique. Our approach aims to mitigate the big dataset's complexity, stemming from class imbalance, size, and high dimensionality, by employing RUS, followed by the integration of the SHAP model within a feature selection framework. The SHAP model is integrated using algorithms including LightGBM, XGBoost, CatBoost, and Decision Tree. To evaluate the effectiveness of our approach, we use the Area Under the Precision-Recall Curve (AUPRC) as the primary evaluation metric to measure the performance of the classification model, utilizing a Random Forest algorithm. Our experiments utilize the Medicare Part D dataset from the Centers for Medicare and Medicaid Services (CMS). Our primary objective is to investigate whether applying RUS with SHAP-based feature selection leads to measurable improvements in binary classification performance. Our findings indicate that the feature subset generated by applying RUS before selecting features using SHAP outperforms those created without the RUS enhancement. This approach not only enhances model performance, but also improves efficiency by reducing computational demands.

As a means of building explainable machine learning models for Big Data, we apply a novel ensemble supervised feature selection technique. The technique is applied to publicly available insurance claims data from the United States public health insurance program, Medicare. We approach Medicare insurance fraud detection as a supervised machine learning task of anomaly detection through the classification of highly imbalanced Big Data. Our objectives for feature selection are to increase efficiency in model training, and to develop more explainable machine learning models for fraud detection. Using two Big Data datasets derived from two different sources of insurance claims data, we demonstrate how our feature selection technique reduces the dimensionality of the datasets by approximately 87.5% without compromising performance. Moreover, the reduction in dimensionality results in machine learning models that are easier to explain, and less prone to overfitting. Therefore, our primary contribution of the exposition of our novel feature selection technique leads to a further contribution to the application domain of automated Medicare insurance fraud detection. We utilize our feature selection technique to provide an explanation of our fraud detection models in terms of the definitions of the selected features. The ensemble supervised feature selection technique we present is flexible in that any collection of machine learning algorithms that maintain a list of feature importance values may be used. Therefore, researchers may easily employ variations of the technique we present.

One-class classifiers hold promise for applications like fraudulent credit card transaction identification. However, interpreting these models to understand which features drive predictions is challenging. Such an understanding is necessary to avoid brute-force approaches to feature selection. This paper explores SHAP (SHapley Additive exPlanations) for feature selection with one-class classffiers on a credit card fraud dataset. We apply SHAP to select key features and evaluate One-Class Gaussian mixture models and One-Class Support Vector Machines. Statistical analysis tests show Gaussian mixture models built with SHAP-selected features perform significantly better than Gaussian mixture models built without feature selection. To the best of our knowledge, we are the first to show the benefit of SHAP-based feature selection to One-Class Gaussian mixture models. Moreover, we show that robust performance with features of the full dataset may be a prerequisite in order for SHAP feature selection to impart further gains. Our results provide novel evidence that SHAP can identify informative features for one-class classifiers.

In Machine Learning , if one class has a significantly larger number of instances (majority) than the other (minority), this condition is defined as class imbalance. With regard to datasets, class imbalance can bias the predictive capabilities of Machine Learning algorithms towards the majority (negative) class, and in situations where false negatives incur a greater penalty than false positives, this imbalance may lead to adverse consequences. Our paper incorporates two case studies, each utilizing a unique approach of three learners (gradient-boosted trees, logistic regression, random forest) and three performance metrics ( Area Under the Receiver Operating Characteristic Curve , Area Under the Precision-Recall Curve , Geometric Mean ) to investigate class rarity in big data. Class rarity, a notably extreme degree of class imbalance, was effected in our experiments by randomly removing minority (positive) instances to artificially generate eight subsets of gradually decreasing positive class instances. All model evaluations were performed through Cross-Validation. In the first case study, which uses a Medicare Part B dataset, performance scores for the learners generally improve with the Area Under the Receiver Operating Characteristic Curve metric as the rarity level decreases, while corresponding scores with the Area Under the Precision-Recall Curve and Geometric Mean metrics show no improvement. In the second case study, which uses a dataset built from Distributed Denial of Service attack attack data (POSTSlowloris Combined), the Area Under the Receiver Operating Characteristic Curve metric produces very high-performance scores for the learners, with all subsets of positive class instances. For the second study, scores for the learners generally improve with the Area Under the Precision-Recall Curve and Geometric Mean metrics as the rarity level decreases. Overall, with regard to both case studies, the Gradient-Boosted Trees (GBT) learner performs the best.

Traditionally, to detect Medicare fraud, a limited number of auditors, or investigators, are responsible for manually inspecting thousands of claims, but only have enough time to look for very specific patterns indicating suspicious behaviors. This chapter provides two case studies to demonstrate the effects of class imbalance with big data on the detection of fraud in the Medicare dataset with List of Excluded Individuals and Entities fraud labels. It utilizes three different datasets, with provider payment and utilization information, released by the Centers for Medicare and Medicaid Services: Medicare Provider Utilization and Payment Data: Physician and Other Supplier, Medicare Provider Utilization and Payment Data: Prescriber, and Medicare Provider Utilization and Payment Data: Referring Durable Medical Equipment, Prosthetics, Orthotics, and Supplies. The chapter considers these parts of Medicare because they cover a wide range of possible provider claims, the information is presented in similar formats, and they are publicly available.

Access to copious amounts of information has reached unprecedented levels, and can generate very large data sources. These big data sources often contain a plethora of useful information but, in some cases, finding what is actually useful can be quite problematic. For binary classification problems, such as fraud detection, a major concern therein is one of class imbalance. This is when a dataset has more of one label versus another, such as a large number of non-fraud observations with comparatively few observations of fraud (which we consider the class of interest). Class rarity further delineates class imbalance with significantly smaller numbers in the class of interest. In this study, we assess the impacts of class rarity in big data, and apply data sampling to mitigate some of the performance degradation caused by rarity. Real-world Medicare claims datasets with known excluded providers are used as fraud labels for a fraud detection scenario, incorporating three machine learning models. We discuss the necessary data processing and engineering steps in order to understand, integrate, and use the Medicare data. From these already imbalanced datasets, we generate three additional datasets representing varying levels of class rarity. We show that, as expected, rarity significantly decreases model performance, but data sampling, specifically random undersampling, can help significantly with rare class detection in identifying Medicare claims fraud cases.

This introduction presents an overview of the key concepts discussed in the subsequent chapters of this book. The book focuses on sampling to reduce the impact of class imbalance on machine learning models. It demonstrates that classification performance across several imbalanced big datasets across different application domains can be significantly improved using Random Undersampling without substantially altering the composition of the original data. The book provides an overview of related works. It describes the Machine Learning (ML) classification algorithms and libraries, to include the evaluation strategy with validation techniques and performance metrics. The book introduces the datasets and how they were processed, model training, and performance evaluation. It involves a real-world Medicare fraud problem, with severe class imbalance. To ease the process of using ML, engineers build the algorithms within software modules or packages, making sure that they work reliably, quickly, and at-scale.

A majority of predictive models should be updated regularly, since the most recent data associated with the model may have a different distribution from that of the original training data. This difference may be critical enough to impact the effectiveness of the machine learning model. In our paper, we investigate the relationship between time and predictive model maintenance. Our work incorporates severely imbalanced big data from three Medicare datasets, namely Part D, DMEPOS, and Combined, that have been used in several fraud detection studies. We build training datasets from year-groupings of 2013, 2014, 2015, 2013–2014, 2014–2015, and 2013–2015. Our test datasets are built from the 2016 data. To mitigate some of the adverse effects from the severe class imbalance in these datasets, the performance of five class ratios obtained by Random Undersampling and five learners is evaluated by the Area Under the Receiver Operating Characteristic Curve metric. The models producing the best values are as follows: Logistic Regression with the 2015 year-grouping at a 99:1 class ratio (Part D); Random Forest with the 2014-2015 year-grouping at a 75:25 class ratio (DMEPOS); and Logistic Regression with the full 2015 year-grouping (Combined). Our experimental results show that the largest training dataset (year-grouping 2013–2015) was not among the selected choices, which indicates that the 2013 data may be outdated. Moreover, we note that because the best model is different for Part D, DMEPOS, and Combined, this suggests that these three datasets may actually be sub-domains requiring unique models within the Medicare fraud detection domain.

The United States healthcare system produces an enormous volume of data with a vast number of financial transactions generated by physicians administering healthcare services. This makes healthcare fraud difficult to detect, especially when there are considerably less fraudulent transactions (documented and readily available) than non-fraudulent. The ability to successfully detect fraudulent activities in healthcare, given such discrepancies, can garner up to $350 billion in recovered monetary losses. In machine learning, when one class has a substantially larger number of instances (majority) compared to the other (minority), this is known as class imbalance. In this paper, we focus specifically on Medicare, utilizing three 'Big Data' Medicare claims datasets with real-world fraudulent physicians. We create a training and test dataset for all three Medicare parts, both separately and combined, to assess fraud detection performance. To emulate class rarity, which indicates particularly severe levels of class imbalance, we generate additional datasets, by removing fraud instances, to determine the effects of rarity on fraud detection performance. Before a machine learning model can be distributed for real-world use, a performance evaluation is necessary to determine the best configuration (e.g. learner, class sampling ratio) and whether the associated error rates are low, indicating good detection rates. With our research, we demonstrate the effects of severe class imbalance and rarity using a training and testing (Train_Test) evaluation method via a hold-out set, and provide our recommendations based on the supervised machine learning results. Additionally, we repeat the same experiments using Cross-Validation, and determine it is a viable substitute for Medicare fraud detection. For machine learning with the severe class imbalance datasets, we found that, as expected, fraud detection performance decreased as the fraudulent instances became more rare. We apply Random Undersampling to both Train_Test and Cross-Validation, for all original and generated datasets, in order to assess potential improvements in fraud detection by reducing the adverse effects of class imbalance and rarity. Overall, our results indicate that the Train_Test method significantly outperforms Cross-Validation.

Evaluating a machine learning model's predictive performance is vital for establishing the practical usability in real-world applications. The use of separate training and test datasets, and cross-validation are common when evaluating machine learning models. The former uses two distinct datasets, whereas cross-validation splits a single dataset into smaller training and test subsets. In real-world production applications, it is critical to establish a model's usefulness by validating it on completely new input data, and not just using the crossvalidation results on a single historical dataset. In this paper, we present results for both evaluation methods, to include performance comparisons. In order to provide meaningful comparative analyses between methods, we perform real-world fraud detection experiments using 2013 to 2016 Medicare durable medical equipment claims data. This Medicare dataset is split into training (2013 to 2015 individual years) and test (2016 only). Using this Medicare case study, we assess the fraud detection performance, across three learners, for both model evaluation methods. We find that using the separate training and test sets generally outperforms cross-validation, indicating a better real-world model performance evaluation. Even so, cross-validation has comparable, but conservative, fraud detection results.

Periodic updating of a machine learning model may become necessary because new data could have a distribution that has drifted significantly over time from the original data distribution, thus impacting the model's usefulness. The primary objective of this paper is to evaluate temporal influence on the maintenance of a predictive model. We investigate the impact of using training data from various year-groupings on a model designed to detect Medicare Part B billing fraud. Training datasets are obtained from year-groupings of 2015, 2014-2015, 2013-2015, and 2012-2015. The test dataset is represented by 2016 data. Our study utilizes five popular learners and five class ratios obtained by Random Undersampling. Using the Area Under the Receiver Operating Characteristic (ROC) Curve as the performance metric, our case study indicates that the Logistic Regression learner yields the highest overall value for the yeargrouping of 2013-2015, with a majority-to-minority ratio of 90:10. For the problem of maintaining predictive models for Medicare fraud, we conclude that a sampled dataset should be chosen over the full dataset and that the largest training dataset (i.e., 2012- 2015) does not always produce the best results. To the best of our knowledge, this is the first big data study that examines the influence of time on the maintenance of machine learning models.

Severe class imbalance between majority and minority classes in Big Data can bias the predictive performance of Machine Learning algorithms toward the majority (negative) class. Where the minority (positive) class holds greater value than the majority (negative) class and the occurrence of false negatives incurs a greater penalty than false positives, the bias may lead to adverse consequences. Our paper incorporates two case studies, each utilizing three learners, six sampling approaches, two performance metrics, and five sampled distribution ratios, to uniquely investigate the effect of severe class imbalance on Big Data analytics. The learners ( Gradient-Boosted Trees, Logistic Regression, Random Forest ) were implemented within the Apache Spark framework. The first case study is based on a Medicare fraud detection dataset. The second case study, unlike the first, includes training data from one source (SlowlorisBig Dataset) and test data from a separate source (POST dataset). Results from the Medicare case study are not conclusive regarding the best sampling approach using Area Under the Receiver Operating Characteristic Curve and Geometric Mean performance metrics. However, it should be noted that the Random Undersampling approach performs adequately in the first case study. For the SlowlorisBig case study, Random Undersampling convincingly outperforms the other five sampling approaches ( Random Oversampling, Synthetic Minority Over-sampling TEchnique, SMOTE-borderline1 , SMOTE-borderline2 , ADAptive SYNthetic ) when measuring performance with Area Under the Receiver Operating Characteristic Curve and Geometric Mean metrics. Based on its classification performance in both case studies, Random Undersampling is the best choice as it results in models with a significantly smaller number of samples, thus reducing computational burden and training time.

Healthcare in the United States is a critical aspect of most people's lives, particularly for the aging demographic. This rising elderly population continues to demand more cost-effective healthcare programs. Medicare is a vital program serving the needs of the elderly in the United States. The growing number of Medicare beneficiaries, along with the enormous volume of money in the healthcare industry, increases the appeal for, and risk of, fraud. In this paper, we focus on the detection of Medicare Part B provider fraud which involves fraudulent activities, such as patient abuse or neglect and billing for services not rendered, perpetrated by providers and other entities who have been excluded from participating in Federal healthcare programs. We discuss Part B data processing and describe a unique process for mapping fraud labels with known fraudulent providers. The labeled big dataset is highly imbalanced with a very limited number of fraud instances. In order to combat this class imbalance, we generate seven class distributions and assess the behavior and fraud detection performance of six different machine learning methods. Our results show that RF100 using a 90:10 class distribution is the best learner with a 0.87302 AUC. Moreover, learner behavior with the 50:50 balanced class distribution is similar to more imbalanced distributions which keep more of the original data. Based on the performance and significance testing results, we posit that retaining more of the majority class information leads to better Medicare Part B fraud detection performance over the balanced datasets across the majority of learners.

Healthcare is an important aspect in everyday life, with quality and affordable care being essential for a population's well-being and life expectancy. Even so, associated costs for medical services continue to rise. One aspect contributing to increased costs in healthcare is waste and fraud. In particular, with the rapidly rising elderly population in the United States, programs like Medicare are subject to high losses due to fraud. Therefore, fraud detection approaches are critical in lessening these losses. Even so, many studies using Medicare data do not provide sufficient details regarding data processing and/or integration making it potentially more difficult to understand the experimental results and challenging to reproduce the experiments. In this paper, we present current research using Medicare data to detect fraud, focusing on data processing and/or integration, and assess any gaps in the provided data-related details. We then present discussions on important details to look for when processing and merging different Medicare datasets indicating opportunities for future work.