Detecting illicit transactions in Anti-Money Laundering (AML) systems remains a significant challenge due to class imbalances and the complexity of financial networks. This study introduces the Multiple Aggregations for Graph Isomorphism Network with Custom Edges (MAGIC) convolution, an enhancement of the Graph Isomorphism Network (GIN) designed to improve the detection of illicit transactions in AML systems. MAGIC integrates edge convolution (GINE Conv) and multiple learnable aggregations, allowing for varied embedding sizes and increased generalization capabilities. Experiments were conducted using synthetic datasets, which simulate real-world transactions, following the experimental setup of previous studies to ensure comparability. MAGIC, when combined with XGBoost as a link predictor, outperformed existing models in 16 out of 24 metrics, with notable improvements in F1 scores and precision. In the most imbalanced dataset, MAGIC achieved an F1 score of 82.6% and a precision of 90.4% for the illicit class. While MAGIC demonstrated high precision, its recall was lower or comparable to the other models, indicating potential areas for future enhancement. Overall, MAGIC presents a robust approach to AML detection, particularly in scenarios where precision and overall quality are critical. Future research should focus on optimizing the model’s recall, potentially by incorporating additional regularization techniques or advanced sampling methods. Additionally, exploring the integration of foundation models like GraphAny could further enhance the model’s applicability in diverse AML environments.
A challenge that data scientists face is building an analytic product that is useful and trustworthy for a given audience. Previously, a set of principles for describing data analyses were defined that can be used to create a data analysis and to characterize the variation between analyses. Here, we introduce a concept called the alignment of a data analysis, which is between the data analyst and an audience. We define an aligned data analysis as the matching of principles between the analyst and the audience for whom the analysis is developed. In this paper, we propose a model for evaluating the alignment of a data analysis and describe some of its properties. We argue that more generally, this framework provides a language for characterizing alignment and can be used as a guide for practicing data scientists to building better data products.
We propose to explore high-dimensional data with categorical outcomes by generalizing the penalized orthogonal-components regression method (POCRE), a supervised dimension reduction method initially proposed for high-dimensional linear regression. This generalized POCRE, i.e., gPOCRE, sequentially builds up orthogonal components by selecting predictors which maximally explain the variation of the response variables. Therefore, gPOCRE simultaneously selects significant predictors and reduces dimensions by constructing linear components of these selected predictors for a high-dimensional generalized linear model. For multiple categorical outcomes, gPOCRE can also construct common components shared by all outcomes to improve the power of selecting variables shared by multiple outcomes. Both simulation studies and real data analysis are carried out to illustrate the performance of gPOCRE.
Forecasting is essential for optimizing resource allocation, particularly during crises such as the unprecedented COVID-19 pandemic. This paper focuses on developing an algorithm for generating k-step-ahead interval forecasts for autoregressive time series. Unlike conventional methods that assume a fixed distribution, our approach utilizes kernel distribution estimation to accommodate the unknown distribution of prediction errors. This flexibility is crucial in real-world data, where deviations from normality are common, and neglecting these deviations can result in inaccurate predictions and unreliable confidence intervals. We evaluate the performance of our method through simulation studies on various autoregressive time series models. The results show that the proposed approach performs robustly, even with small sample sizes, as low as 50 observations. Moreover, our method outperforms traditional linear model-based prediction intervals and those derived from the empirical distribution function, particularly when the underlying data distribution is non-normal. This highlights the algorithm’s flexibility and accuracy for interval forecasting in non-Gaussian contexts. We also apply the method to log-transformed weekly COVID-19 case counts from lower-middle-income countries, covering the period from June 1, 2020, to March 13, 2022.
Approximately 15% of adults in the United States (U.S.) are afflicted with chronic kidney disease (CKD). For CKD patients, the progressive decline of kidney function is intricately related to hospitalizations due to cardiovascular disease and eventual “terminal” events, such as kidney failure and mortality. To unravel the mechanisms underlying the disease dynamics of these interdependent processes, including identifying influential risk factors, as well as tailoring decision-making to individual patient needs, we develop a novel Bayesian multivariate joint model for the intercorrelated outcomes of kidney function (as measured by longitudinal estimated glomerular filtration rate), recurrent cardiovascular events, and competing-risk terminal events of kidney failure and death. The proposed joint modeling approach not only facilitates the exploration of risk factors associated with each outcome, but also allows dynamic updates of cumulative incidence probabilities for each competing risk for future subjects based on their basic characteristics and a combined history of longitudinal measurements and recurrent events. We propose efficient and flexible estimation and prediction procedures within a Bayesian framework employing Markov Chain Monte Carlo methods. The predictive performance of our model is assessed through dynamic area under the receiver operating characteristic curves and the expected Brier score. We demonstrate the efficacy of the proposed methodology through extensive simulations. Proposed methodology is applied to data from the Chronic Renal Insufficiency Cohort study established by the National Institute of Diabetes and Digestive and Kidney Diseases to address the rising epidemic of CKD in the U.S.
Extensive literature has been proposed for the analysis of correlated survival data. Subjects within a cluster share some common characteristics, e.g., genetic and environmental factors, so their time-to-event outcomes are correlated. The frailty model under proportional hazards assumption has been widely applied for the analysis of clustered survival outcomes. However, the prediction performance of this method can be less satisfactory when the risk factors have complicated effects, e.g., nonlinear and interactive. To deal with these issues, we propose a neural network frailty Cox model that replaces the linear risk function with the output of a feed-forward neural network. The estimation is based on quasi-likelihood using Laplace approximation. A simulation study suggests that the proposed method has the best performance compared with existing methods. The method is applied to the clustered time-to-failure prediction within the kidney transplantation facility using the national kidney transplant registry data from the U.S. Organ Procurement and Transplantation Network. All computer programs are available at https://github.com/rivenzhou/deep_learning_clustered.
Cellular deconvolution is a key approach to deciphering the complex cellular makeup of tissues by inferring the composition of cell types from bulk data. Traditionally, deconvolution methods have focused on a single molecular modality, relying either on RNA sequencing (RNA-seq) to capture gene expression or on DNA methylation (DNAm) to reveal epigenetic profiles. While these single-modality approaches have provided important insights, they often lack the depth needed to fully understand the intricacies of cellular compositions, especially in complex tissues. To address these limitations, we introduce EMixed, a versatile framework designed for both single-modality and multi-omics cellular deconvolution. EMixed models raw RNA counts and DNAm counts or frequencies via allocation models that assign RNA transcripts and DNAm reads to cell types, and uses an expectation-maximization (EM) algorithm to estimate parameters. Benchmarking results demonstrate that EMixed significantly outperforms existing methods across both single-modality and multi-modality applications, underscoring the broad utility of this approach in enhancing our understanding of cellular heterogeneity.
Loan behavior modeling is crucial in financial engineering. In particular, predicting loan prepayment based on large-scale historical time series data of massive customers is challenging. Existing approaches, such as logistic regression or nonparametric regression, could only model the direct relationship between the features and the prepayments. Motivated by extracting the hidden states of loan behavior, we propose the smoothing spline state space (QuadS) model based on a hidden Markov model with varying transition and emission matrices modeled by smoothing splines. In contrast to existing methods, our method benefits from capturing the loans’ unobserved state transitions, which not only increases prediction performances but also provides more interpretability. The overall model is learned by EM algorithm iterations, and within each iteration, smoothing splines are fitted with penalized least squares. Simulation studies demonstrate the effectiveness of the proposed method. Furthermore, a real-world case study using loan data from the Federal National Mortgage Association illustrates the practical applicability of our model. The QuadS model not only provides reliable predictions but also uncovers meaningful, hidden behavior patterns that can offer valuable insights for the financial industry.
Heart rate data collected from wearable devices – one type of time series data – could provide insights into activities, stress levels, and health. Yet, consecutive missing segments (i.e., gaps) that commonly occur due to improper device placement or device malfunction could distort the temporal patterns inherent in the data and undermine the validity of downstream analyses. This study proposes an innovative iterative procedure to fill gaps in time series data that capitalizes on the denoising capability of Singular Spectrum Analysis (SSA) and eliminates SSA’s requirement of pre-specifying the window length and number of groups. The results of simulations demonstrate that the performance of SSA-based gap-filling methods depends on the choice of window length, number of groups, and the percentage of missing values. In contrast, the proposed method consistently achieves the lowest rates of reconstruction error and gap-filling error across a variety of combinations of the factors manipulated in the simulations. The simulation findings also highlight that the commonly recommended long window length – half of the time series length – may not apply to time series with varying frequencies such as heart rate data. The initialization step of the proposed method that involves a large window length and the first four singular values in the iterative singular value decomposition process not only avoids convergence issues but also facilitates imputation accuracy in subsequent iterations. The proposed method provides the flexibility for researchers to conduct gap-filling solely or in combination with denoising on time series data and thus widens the applications.
In many medical comparative studies, subjects may provide either bilateral or unilateral data. While numerous testing procedures have been proposed for bilateral data that account for the intra-class correlation between paired organs of the same individual, few studies have thoroughly explored combined correlated bilateral and unilateral data. Ma and Wang (2021) introduced three test procedures based on the maximum likelihood estimation (MLE) algorithm for general g groups. In this article, we employ a model-based approach that treats the measurements from both eyes of each subject as repeated observations. We then compare this approach with Ma and Wang’s Score test procedure. Monte Carlo simulations demonstrate that the MLE-based Score test offers certain advantages under specific conditions. However, this model-based method lacks an explicit form for the test statistic, limiting its potential for further development of an exact test.