Propensity score estimation: neural networks, support vector machines, decision trees (CART), and meta-classifiers as alternatives to logistic regression

Association between comorbidities at ICU admission and post-Sepsis physical impairment: A retrospective cohort study

PURPOSE

Few studies have measured the association between pre-existing comorbidities and post-sepsis physical impairment. The study aimed to estimate the risk of physical impairment at hospital discharge among sepsis patients, adjusting for pre-existing physical impairment prior to ICU admission and in-hospital mortality.

MATERIALS AND METHODS

We analyzed all consecutive adult patients admitted to an ICU in a tertiary community hospital, Kameda Medical Center, with sepsis diagnosis from September 2014 to October 2020. Inverse probability attrition weighting using machine learning was employed to estimate the risk of physical impairment at hospital discharge for sepsis patients with and without pre-existing comorbidities at ICU admission. This estimation was adjusted for baseline covariates, pre-ICU physical impairment, and in-hospital mortality.

RESULTS

Of 889 sepsis patients analyzed, 668 [75.1%] had at least one comorbidity and 221 [24.9%] had no comorbidities at ICU admission. Upon adjusting for baseline covariates, pre-ICU physical impairment, and in-hospital mortality, pre-existing comorbidities were not associated with an elevated risk of physical impairment at hospital discharge (RR: 1.02, 95% CI: 0.92, 1.14).

CONCLUSIONS

Pre-existing comorbidities prior to ICU admission were not associated with an increased risk of physical impairment at hospital discharge among sepsis patients after adjusting for baseline covariates and in-hospital mortality.

Predictive Modeling for One-Year Lower Extremity Endovascular Revascularization Failure in Black Persons

INTRODUCTION

Black persons bear a disproportionate burden of peripheral artery disease (PAD) and experience higher rates of endovascular revascularization failure (ERF) when compared with non-Hispanic White persons. We aimed to identify predictors of ERF in Black persons using predictive modeling.

METHODS

This retrospective study included all persons identifying as Black who underwent an initial endovascular revascularization procedure for PAD between 2011 and 2018 at a midwestern tertiary care center. Three predictive models were developed using (1) logistic regression, (2) penalized logistic regression (least absolute shrinkage and selection operator [LASSO]), and (3) random forest (RF). Predictive performance was evaluated under repeated cross-validation.

RESULTS

Of the 163 individuals included in the study, 113 (63.1%) experienced ERF at 1 y. Those with ERF had significant differences in symptom status (P < 0.001), lesion location (P < 0.001), diabetes status (P = 0.037), and annual procedural volume of the attending surgeon (P < 0.001). Logistic regression and LASSO models identified tissue loss, smoking, femoro-popliteal lesion location, and diabetes control as risk factors for ERF. The RF model identified annual procedural volume, age, PAD symptoms, number of comorbidities, and lesion location as most predictive variables. LASSO and RF models were more sensitive than logistic regression but less specific, although all three methods had an overall accuracy of ≥75%.

CONCLUSIONS

Black persons undergoing endovascular revascularization for PAD are at high risk of ERF, necessitating need for targeted intervention. Predictive models may be clinically useful for identifying high-risk patients, although individual predictors of ERF varied by model. Further exploration into these models may improve limb salvage for this population.

A fast bootstrap algorithm for causal inference with large data

Estimating causal effects from large experimental and observational data has become increasingly prevalent in both industry and research. The bootstrap is an intuitive and powerful technique used to construct standard errors and confidence intervals of estimators. Its application however can be prohibitively demanding in settings involving large data. In addition, modern causal inference estimators based on machine learning and optimization techniques exacerbate the computational burden of the bootstrap. The bag of little bootstraps has been proposed in non-causal settings for large data but has not yet been applied to evaluate the properties of estimators of causal effects. In this article, we introduce a new bootstrap algorithm called causal bag of little bootstraps for causal inference with large data. The new algorithm significantly improves the computational efficiency of the traditional bootstrap while providing consistent estimates and desirable confidence interval coverage. We describe its properties, provide practical considerations, and evaluate the performance of the proposed algorithm in terms of bias, coverage of the true 95% confidence intervals, and computational time in a simulation study. We apply it in the evaluation of the effect of hormone therapy on the average time to coronary heart disease using a large observational data set from the Women's Health Initiative.

Improving Inverse Probability Weighting by Post-calibrating Its Propensity Scores

Theoretical guarantees for causal inference using propensity scores are partially based on the scores behaving like conditional probabilities. However, prediction scores between zero and one do not necessarily behave like probabilities, especially when output by flexible statistical estimators. We perform a simulation study to assess the error in estimating the average treatment effect before and after applying a simple and well-established postprocessing method to calibrate the propensity scores. We observe that postcalibration reduces the error in effect estimation and that larger improvements in calibration result in larger improvements in effect estimation. Specifically, we find that expressive tree-based estimators, which are often less calibrated than logistic regression-based models initially, tend to show larger improvements relative to logistic regression-based models. Given the improvement in effect estimation and that postcalibration is computationally cheap, we recommend its adoption when modeling propensity scores with expressive models.

Investigating disparities in smoking cessation treatment for veterans with multiple sclerosis: A national analysis

BACKGROUND AND AIMS

Smoking is a risk factor for multiple sclerosis (MS) development, symptom burden, decreased medication efficacy, and increased disease-related mortality. Veterans with MS (VwMS) smoke at critically high rates; however, treatment rates and possible disparities are unknown. To promote equitable treatment, we aim to investigate smoking cessation prescription practices for VwMS across social determinant factors.

METHODS

We extracted data from the national Veterans Health Administration electronic health records between October 1, 2017, and September 30, 2018. To derive marginal estimates of the association of MS with receipt of smoking-cessation pharmacotherapy, we used propensity score matching through the extreme gradient boosting machine learning model. VwMS who smoke were matched with veterans without MS who smoke on factors including age, race, depression, and healthcare visits. To assess the marginal association of MS with different cessation treatments, we used logistic regression and conducted stratified analyses by sex, race, and ethnicity.

RESULTS

The matched sample achieved a good balance across most covariates, compared to the pre-match sample. VwMS (n = 3320) had decreased odds of receiving prescriptions for nicotine patches ([Odds Ratio]OR = 0.86, p < .01), non-patch nicotine replacement therapy (NRT; OR = 0.81, p < .001), and standard practice dual NRT (OR = 0.77, p < .01), compared to matches without MS (n = 13,280). Men with MS had lower odds of receiving prescriptions for nicotine patches (OR = 0.88, p = .05), non-patch NRT (OR = 0.77, p < .001), and dual NRT (OR = 0.72, p < .001). Similarly, Black VwMS had lower odds of receiving prescriptions for patches (OR = 0.62, p < .001), non-patch NRT (OR = 0.75, p < .05), and dual NRT (OR = 0.52, p < .01). The odds of receiving prescriptions for bupropion or varenicline did not differ between VwMS and matches without MS.

CONCLUSION

VwMS received significantly less smoking cessation treatment, compared to matched controls without MS, showing a critical gap in health services as VwMS are not receiving dual NRT as the standard of care. Prescription rates were especially lower for male and Black VwMS, suggesting that under-represented demographic groups outside of the white female category, most often considered as the "traditional MS" group, could be under-treated regarding smoking cessation support. This foundational work will help inform future work to promote equitable treatment and implementation of cessation interventions for people living with MS.

Noninvasive Risk Prediction Models for Heart Failure Using Proportional Jaccard Indices and Comorbidity Patterns

Background

In the post-coronavirus disease 2019 (COVID-19) era, remote diagnosis and precision preventive medicine have emerged as pivotal clinical medicine applications. This study aims to develop a digital health-monitoring tool that utilizes electronic medical records (EMRs) as the foundation for performing a non-random correlation analysis among different comorbidity patterns for heart failure (HF).

Methods

Novel similarity indices, including proportional Jaccard index (PJI), multiplication of the odds ratio proportional Jaccard index (OPJI), and alpha proportional Jaccard index (APJI), provide a fundamental framework for constructing machine learning models to predict the risk conditions associated with HF.

Results

Our models were constructed for different age groups and sexes and yielded accurate predictions of high-risk HF across demographics. The results indicated that the optimal prediction model achieved a notable accuracy of 82.1% and an area under the curve (AUC) of 0.878.

Conclusions

Our noninvasive HF risk prediction system is based on historical EMRs and provides a practical approach. The proposed indices provided simple and straightforward comparative indicators of comorbidity pattern matching within individual EMRs. All source codes developed for our noninvasive prediction models can be retrieved from GitHub.

Intelligent decision support systems for dementia care: A scoping review

In the context of dementia care, Artificial Intelligence (AI) powered clinical decision support systems have the potential to enhance diagnosis and management. However, the scope and challenges of applying these technologies remain unclear. This scoping review aims to investigate the current state of AI applications in the development of intelligent decision support systems for dementia care. We conducted a comprehensive scoping review of empirical studies that utilised AI-powered clinical decision support systems in dementia care. The results indicate that AI applications in dementia care primarily focus on diagnosis, with limited attention to other aspects outlined in the World Health Organization (WHO) Global Action Plan on the Public Health Response to Dementia 2017-2025 (GAPD). A trifecta of challenges, encompassing data availability, cost considerations, and AI algorithm performance, emerges as noteworthy barriers in adoption of AI applications in dementia care. To address these challenges and enhance AI reliability, we propose a novel approach: a digital twin-based patient journey model. Future research should address identified gaps in GAPD action areas, navigate data-related obstacles, and explore the implementation of digital twins. Additionally, it is imperative to emphasize that addressing trust and combating the stigma associated with AI in healthcare should be a central focus of future research directions.

Posttraumatic Stress Disorder Is Associated With Elevated Risk of Incident Stroke and Transient Ischemic Attack in Women Veterans

BACKGROUND

Posttraumatic stress disorder (PTSD) has been associated with ischemic heart disease in women veterans, but evidence for associations with other cardiovascular disorders remains limited in this population. This retrospective longitudinal cohort study evaluated the association of PTSD with incident stroke/transient ischemic attack (TIA) in women veterans.

METHODS AND RESULTS

Veterans Health Administration electronic health records were used to identify women veterans aged ≥18 years engaged with Veterans Health Administration health care from January 1, 2000 to December 31, 2019. We identified women veterans with and without PTSD without a history of stroke or TIA at start of follow-up. Propensity score matching was used to match groups on age, race or ethnicity, traditional cardiovascular risk factors, female-specific risk factors, a range of mental and physical health conditions, and number of prior health care visits. PTSD, stroke, TIA, and risk factors used in propensity score matching were based on diagnostic codes. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% CIs for associations of PTSD with an incident stroke/TIA composite. Subanalyses considered stroke and TIA separately, plus age- and race- or ethnicity-stratified analyses were carried out. The analytic sample included 208 092 women veterans (104 046 with and 104 046 without PTSD). PTSD was associated with a greater rate of developing stroke/TIA (HR, 1.33 [95% CI, 1.25-1.42], P<0.001). This elevated risk was especially pronounced in women <50 years old and in Hispanic/Latina women.

CONCLUSIONS

Findings indicate a strong association of PTSD with incident stroke/TIA in women veterans. Research is needed to determine whether addressing PTSD and its downstream consequences can offset this risk.

Propensity score stratified MAP prior and posterior inference for incorporating information across multiple potentially heterogeneous data sources

Incorporation of external information is becoming increasingly common when designing clinical trials. Availability of multiple sources of information has inspired the development of methodologies that account for potential heterogeneity not only between the prospective trial and the pooled external data sources but also between the different external data sources themselves. Our approach proposes an intuitive way of handling such a scenario for the continuous outcomes setting by using propensity score-based stratification and then utilizing robust meta-analytic predictive priors for each stratum to incorporate the prior data to distinguish among different external data sources in each stratum. Through extensive simulations, our approach proves to be more efficient and less biased than the currently available methods. A real case study using clinical trials that study schizophrenia from multiple different sources is also included.

Impact of Artificial Intelligence (AI) Technology in Healthcare Sector: A Critical Evaluation of Both Sides of the Coin

The influence of artificial intelligence (AI) has drastically risen in recent years, especially in the field of medicine. Its influence has spread so greatly that it is determined to become a pillar in the future medical world. A comprehensive literature search related to AI in healthcare was performed in the PubMed database and retrieved the relevant information from suitable ones. AI excels in aspects such as rapid adaptation, high diagnostic accuracy, and data management that can help improve workforce productivity. With this potential in sight, the FDA has continuously approved more machine learning (ML) software to be used by medical workers and scientists. However, there are few controversies such as increased chances of data breaches, concern for clinical implementation, and potential healthcare dilemmas. In this article, the positive and negative aspects of AI implementation in healthcare are discussed, as well as recommended some potential solutions to the potential issues at hand.

Combining Super Learner with high-dimensional propensity score to improve confounding adjustment: A real-world application in chronic lymphocytic leukemia

PURPOSE

High-dimensional propensity score (hdPS) is a semiautomated method that leverages a vast number of covariates available in healthcare databases to improve confounding adjustment. A novel combined Super Learner (SL)-hdPS approach was proposed to assist with selecting the number of covariates for propensity score inclusion, and was found in plasmode simulation studies to improve bias reduction and precision compared to hdPS alone. However, the approach has not been examined in the applied setting.

METHODS

We compared SL-hdPS's performance with that of several hdPS models, each with prespecified covariates and a different number of empirically-identified covariates, using a cohort study comparing real-world bleeding rates between ibrutinib- and bendamustine-rituximab (BR)-treated individuals with chronic lymphocytic leukemia in Optum's de-identified Clinformatics® Data Mart commercial claims database (2013-2020). We used inverse probability of treatment weighting for confounding adjustment and Cox proportional hazards regression to estimate hazard ratios (HRs) for bleeding outcomes. Parameters of interest included prespecified and empirically-identified covariate balance (absolute standardized difference [ASD] thresholds of <0.10 and <0.05) and outcome HR precision (95% confidence intervals).

RESULTS

We identified 2423 ibrutinib- and 1102 BR-treated individuals. Including >200 empirically-identified covariates in the hdPS model compromised covariate balance at both ASD thresholds. SL-hdPS balanced more covariates than all individual hdPS models at both ASD thresholds. The bleeding HR 95% confidence intervals were generally narrower with SL-hdPS than with individual hdPS models.

CONCLUSION

In a real-world application, hdPS was sensitive to the number of covariates included, while use of SL for covariate selection resulted in improved covariate balance and possibly improved precision.

RKHS-based covariate balancing for survival causal effect estimation

Survival causal effect estimation based on right-censored data is of key interest in both survival analysis and causal inference. Propensity score weighting is one of the most popular methods in the literature. However, since it involves the inverse of propensity score estimates, its practical performance may be very unstable, especially when the covariate overlap is limited between treatment and control groups. To address this problem, a covariate balancing method is developed in this paper to estimate the counterfactual survival function. The proposed method is nonparametric and balances covariates in a reproducing kernel Hilbert space (RKHS) via weights that are counterparts of inverse propensity scores. The uniform rate of convergence for the proposed estimator is shown to be the same as that for the classical Kaplan-Meier estimator. The appealing practical performance of the proposed method is demonstrated by a simulation study as well as two real data applications to study the causal effect of smoking on survival time of stroke patients and that of endotoxin on survival time for female patients with lung cancer respectively.

Leveraging baseline covariates to analyze small cluster-randomized trials with a rare binary outcome

Cluster-randomized trials (CRTs) involve randomizing entire groups of participants-called clusters-to treatment arms but are often comprised of a limited or fixed number of available clusters. While covariate adjustment can account for chance imbalances between treatment arms and increase statistical efficiency in individually randomized trials, analytical methods for individual-level covariate adjustment in small CRTs have received little attention to date. In this paper, we systematically investigate, through extensive simulations, the operating characteristics of propensity score weighting and multivariable regression as two individual-level covariate adjustment strategies for estimating the participant-average causal effect in small CRTs with a rare binary outcome and identify scenarios where each adjustment strategy has a relative efficiency advantage over the other to make practical recommendations. We also examine the finite-sample performance of the bias-corrected sandwich variance estimators associated with propensity score weighting and multivariable regression for quantifying the uncertainty in estimating the participant-average treatment effect. To illustrate the methods for individual-level covariate adjustment, we reanalyze a recent CRT testing a sedation protocol in 31 pediatric intensive care units.

High-throughput target trial emulation for Alzheimer's disease drug repurposing with real-world data

Target trial emulation is the process of mimicking target randomized trials using real-world data, where effective confounding control for unbiased treatment effect estimation remains a main challenge. Although various approaches have been proposed for this challenge, a systematic evaluation is still lacking. Here we emulated trials for thousands of medications from two large-scale real-world data warehouses, covering over 10 years of clinical records for over 170 million patients, aiming to identify new indications of approved drugs for Alzheimer's disease. We assessed different propensity score models under the inverse probability of treatment weighting framework and suggested a model selection strategy for improved baseline covariate balancing. We also found that the deep learning-based propensity score model did not necessarily outperform logistic regression-based methods in covariate balancing. Finally, we highlighted five top-ranked drugs (pantoprazole, gabapentin, atorvastatin, fluticasone, and omeprazole) originally intended for other indications with potential benefits for Alzheimer's patients.

Performance of modeling and balancing approach methods when using weights to estimate treatment effects in observational time-to-event settings

In observational studies weighting techniques are often used to overcome bias due to confounding. Modeling approaches, such as inverse propensity score weighting, are popular, but often rely on the correct specification of a parametric model wherein neither balance nor stability are targeted. More recently, balancing approach methods that directly target covariate imbalances have been proposed, and these allow the researcher to explicitly set the desired balance constraints. In this study, we evaluate the finite sample properties of different modeling and balancing approach methods, when estimating the marginal hazard ratio, through Monte Carlo simulations. The use of the different methods is also illustrated by analyzing data from the Swedish stroke register to estimate the effect of prescribing oral anticoagulants on time to recurrent stroke or death in stroke patients with atrial fibrillation. In simulated scenarios with good overlap and low or no model misspecification the balancing approach methods performed similarly to the modeling approach methods. In scenarios with bad overlap and model misspecification, the modeling approach method incorporating variable selection performed better than the other methods. The results indicate that it is valuable to use methods that target covariate balance when estimating marginal hazard ratios, but this does not in itself guarantee good performance in situations with, e.g., poor overlap, high censoring, or misspecified models/balance constraints.

Discriminative machine learning for maximal representative subsampling

Biased population samples pose a prevalent problem in the social sciences. Therefore, we present two novel methods that are based on positive-unlabeled learning to mitigate bias. Both methods leverage auxiliary information from a representative data set and train machine learning classifiers to determine the sample weights. The first method, named maximum representative subsampling (MRS), uses a classifier to iteratively remove instances, by assigning a sample weight of 0, from the biased data set until it aligns with the representative one. The second method is a variant of MRS - Soft-MRS - that iteratively adapts sample weights instead of removing samples completely. To assess the effectiveness of our approach, we induced artificial bias in a public census data set and examined the corrected estimates. We compare the performance of our methods against existing techniques, evaluating the ability of sample weights created with Soft-MRS or MRS to minimize differences and improve downstream classification tasks. Lastly, we demonstrate the applicability of the proposed methods in a real-world study of resilience research, exploring the influence of resilience on voting behavior. Through our work, we address the issue of bias in social science, amongst others, and provide a versatile methodology for bias reduction based on machine learning. Based on our experiments, we recommend to use MRS for downstream classification tasks and Soft-MRS for downstream tasks where the relative bias of the dependent variable is relevant.

Abdul Hamid AA, Zainuddin NH, Aruchunan E, Abdullah MT. Enhancing COVID-19 Classification Accuracy with a Hybrid SVM-LR Model

Support ector achine (SVM) is a newer machine learning algorithm for classification, while logistic regression (LR) is an older statistical classification method. Despite the numerous studies contrasting SVM and LR, new improvements such as bagging and ensemble have been applied to them since these comparisons were made. This study proposes a new hybrid model based on SVM and LR for predicting small events per variable (EPV). The performance of the hybrid, SVM, and LR models with different EPV values was evaluated using COVID-19 data from December 2019 to May 2020 provided by the WHO. The study found that the hybrid model had better classification performance than SVM and LR in terms of accuracy, mean squared error (MSE), and root mean squared error (RMSE) for different EPV values. This hybrid model is particularly important for medical authorities and practitioners working in the face of future pandemics.

An improved multiply robust estimator for the average treatment effect

BACKGROUND

In observational studies, double robust or multiply robust (MR) approaches provide more protection from model misspecification than the inverse probability weighting and g-computation for estimating the average treatment effect (ATE). However, the approaches are based on parametric models, leading to biased estimates when all models are incorrectly specified. Nonparametric methods, such as machine learning or nonparametric double robust approaches, are robust to model misspecification, but the efficiency of nonparametric methods is low.

METHOD

In the study, we proposed an improved MR method combining parametric and nonparametric models based on the previous MR method (Han, JASA 109(507):1159-73, 2014) to improve the robustness to model misspecification and the efficiency. We performed comprehensive simulations to evaluate the performance of the proposed method.

RESULTS

Our simulation study showed that the MR estimators with only outcome regression (OR) models, where one of the models was a nonparametric model, were the most recommended because of the robustness to model misspecification and the lowest root mean square error (RMSE) when including a correct parametric OR model. And the performance of the recommended estimators was comparative, even if all parametric models were misspecified. As an application, the proposed method was used to estimate the effect of social activity on depression levels in the China Health and Retirement Longitudinal Study dataset.

CONCLUSIONS

The proposed estimator with nonparametric and parametric models is more robust to model misspecification.

Biomarker discovery process at binomial decision point (2BDP): Analytical pipeline to construct biomarker panel

A clinical incident is typically manifested by several molecular events; therefore, it seems logical that a successful diagnosis, prognosis, or stratification of a clinical landmark require multiple biomarkers. In this report, we presented a machine learning pipeline, namely "Biomarker discovery process at binomial decision point" (2BDP) that took an integrative approach in systematically curating independent variables (e.g., multiple molecular markers) to explain an output variable (e.g., clinical landmark) of binary in nature. In a logical sequence, 2BDP includes feature selection, unsupervised model development and cross validation. In the present work, the efficiency of 2BDP was demonstrated by finding three biomarker panels that independently explained three stages of Alzheimer's disease (AD) marked as Braak stages I, II and III, respectively. We designed three assortments from the entire cohort based on these Braak stages; subsequently, each assortment was split into two populations at Braak score I, II or III. 2BDP systematically integrated random forest and logistic regression fitting model to find biomarker panels with minimum features that explained these three assortments, e.g., significantly differentiated two populations segregated by Braak stage I, II or III, respectively. Thereafter, the efficacies of these panels were measured by the area under the curve (AUC) values of the receiver operating characteristic (ROC) plot. The AUC-ROC was calculated by two cross-validation methods. Final set of gene markers was a mix of novel and a priori established AD signatures. These markers were weighted by unique coefficients and linearly connected in a group of 2-10 to explain Braak stage I, II or III by AUC ≥ 0.8. Small sample size and a lack of distinctly recruited Training and Test sets were the limitations of the present undertaking; yet 2BDP demonstrated its capability to curate a panel of optimum numbers of biomarkers to describe the outcome variable with high efficacy.

Directed acyclic graphs for clinical research: a tutorial

Directed acyclic graphs (DAGs) are useful tools for visualizing the hypothesized causal structures in an intuitive way and selecting relevant confounders in causal inference. However, in spite of their increasing use in clinical and surgical research, the causal graphs might also be misused by a lack of understanding of the central principles. In this article, we aim to introduce the basic terminology and fundamental rules of DAGs, and DAGitty, a user-friendly program that easily displays DAGs. Specifically, we describe how to determine variables that should or should not be adjusted based on the backdoor criterion with examples. In addition, the occurrence of the various types of biases is discussed with caveats, including the problem caused by the traditional approach using p-values for confounder selection. Moreover, a detailed guide to DAGitty is provided with practical examples regarding minimally invasive surgery. Essentially, the primary benefit of DAGs is to aid researchers in clarifying the research questions and the corresponding designs based on the domain knowledge. With these strengths, we propose that the use of DAGs may contribute to rigorous research designs, and lead to transparency and reproducibility in research on minimally invasive surgery.

Reweighting estimators to extend the external validity of clinical trials: methodological considerations

Methods to extend the strong internal validity of randomized controlled trials to reliably estimate treatment effects in target populations are gaining attention. This paper enumerates steps recommended for undertaking such extended inference, discusses currently viable choices for each one, and provides recommendations. We demonstrate a complete extended inference from a clinical trial studying a pharmaceutical treatment for Alzheimer's disease (AD) to a realistic target population of European residents diagnosed with AD. This case study highlights approaches to overcoming practical difficulties and demonstrates limitations of reliably extending inference from a trial to a real-world population.

Genetic matching for time-dependent treatments: a longitudinal extension and simulation study

BACKGROUND

Longitudinal matching can mitigate confounding in observational, real-world studies of time-dependent treatments. To date, these methods have required iterative, manual re-specifications to achieve covariate balance. We propose a longitudinal extension of genetic matching, a machine learning approach that automates balancing of covariate histories. We examine performance by comparing the proposed extension against baseline propensity score matching and time-dependent propensity score matching.

METHODS

To evaluate comparative performance, we developed a Monte Carlo simulation framework that reflects a static treatment assigned at multiple time points. Data generation considers a treatment assignment model, a continuous outcome model, and underlying covariates. In simulation, we generated 1,000 datasets, each consisting of 1,000 subjects, and applied: (1) nearest neighbour matching on time-invariant, baseline propensity scores; (2) sequential risk set matching on time-dependent propensity scores; and (3) longitudinal genetic matching on time-dependent covariates. To measure comparative performance, we estimated covariate balance, efficiency, bias, and root mean squared error (RMSE) of treatment effect estimates. In scenario analysis, we varied underlying assumptions for assumed covariate distributions, correlations, treatment assignment models, and outcome models.

RESULTS

In all scenarios, baseline propensity score matching resulted in biased effect estimation in the presence of time-dependent confounding, with mean bias ranging from 29.7% to 37.2%. In contrast, time-dependent propensity score matching and longitudinal genetic matching achieved stronger covariate balance and yielded less biased estimation, with mean bias ranging from 0.7% to 13.7%. Across scenarios, longitudinal genetic matching achieved similar or better performance than time-dependent propensity score matching without requiring manual re-specifications or normality of covariates.

CONCLUSIONS

While the most appropriate longitudinal method will depend on research questions and underlying data patterns, our study can help guide these decisions. Simulation results demonstrate the validity of our longitudinal genetic matching approach for supporting future real-world assessments of treatments accessible at multiple time points.

Fixed Effects or Mixed Effects Classifiers? Evidence From Simulated and Archival Data

This study seeks to compare fixed and mixed effects models for the purposes of predictive classification in the presence of multilevel data. The first part of the study utilizes a Monte Carlo simulation to compare fixed and mixed effects logistic regression and random forests. An applied examination of the prediction of student retention in the public-use U.S. PISA data set was considered to verify the simulation findings. Results of this study indicate fixed effects models performed comparably with mixed effects models across both the simulation and PISA examinations. Results broadly suggest that researchers should be cognizant of the type of predictors and data structure being used, as these factors carried more weight than did the model type.

Machine learning models for diagnosis and prognosis of Parkinson's disease using brain imaging: general overview, main challenges, and future directions

Parkinson's disease (PD) is a progressive and complex neurodegenerative disorder associated with age that affects motor and cognitive functions. As there is currently no cure, early diagnosis and accurate prognosis are essential to increase the effectiveness of treatment and control its symptoms. Medical imaging, specifically magnetic resonance imaging (MRI), has emerged as a valuable tool for developing support systems to assist in diagnosis and prognosis. The current literature aims to improve understanding of the disease's structural and functional manifestations in the brain. By applying artificial intelligence to neuroimaging, such as deep learning (DL) and other machine learning (ML) techniques, previously unknown relationships and patterns can be revealed in this high-dimensional data. However, several issues must be addressed before these solutions can be safely integrated into clinical practice. This review provides a comprehensive overview of recent ML techniques analyzed for the automatic diagnosis and prognosis of PD in brain MRI. The main challenges in applying ML to medical diagnosis and its implications for PD are also addressed, including current limitations for safe translation into hospitals. These challenges are analyzed at three levels: disease-specific, task-specific, and technology-specific. Finally, potential future directions for each challenge and future perspectives are discussed.

Artificial Intelligence in Pharmaceutical Technology and Drug Delivery Design

Artificial intelligence (AI) has emerged as a powerful tool that harnesses anthropomorphic knowledge and provides expedited solutions to complex challenges. Remarkable advancements in AI technology and machine learning present a transformative opportunity in the drug discovery, formulation, and testing of pharmaceutical dosage forms. By utilizing AI algorithms that analyze extensive biological data, including genomics and proteomics, researchers can identify disease-associated targets and predict their interactions with potential drug candidates. This enables a more efficient and targeted approach to drug discovery, thereby increasing the likelihood of successful drug approvals. Furthermore, AI can contribute to reducing development costs by optimizing research and development processes. Machine learning algorithms assist in experimental design and can predict the pharmacokinetics and toxicity of drug candidates. This capability enables the prioritization and optimization of lead compounds, reducing the need for extensive and costly animal testing. Personalized medicine approaches can be facilitated through AI algorithms that analyze real-world patient data, leading to more effective treatment outcomes and improved patient adherence. This comprehensive review explores the wide-ranging applications of AI in drug discovery, drug delivery dosage form designs, process optimization, testing, and pharmacokinetics/pharmacodynamics (PK/PD) studies. This review provides an overview of various AI-based approaches utilized in pharmaceutical technology, highlighting their benefits and drawbacks. Nevertheless, the continued investment in and exploration of AI in the pharmaceutical industry offer exciting prospects for enhancing drug development processes and patient care.

Recent Developments in Causal Inference and Machine Learning

This article reviews recent advances in causal inference relevant to sociology. We focus on a selective subset of contributions aligning with four broad topics: causal effect identification and estimation in general, causal effect heterogeneity, causal effect mediation, and temporal and spatial interference. We describe how machine learning, as an estimation strategy, can be effectively combined with causal inference, which has been traditionally concerned with identification. The incorporation of machine learning in causal inference enables researchers to better address potential biases in estimating causal effects and uncover heterogeneous causal effects. Uncovering sources of effect heterogeneity is key for generalizing to populations beyond those under study. While sociology has long emphasized the importance of causal mechanisms, historical and life-cycle variation, and social contexts involving network interactions, recent conceptual and computational advances facilitate more principled estimation of causal effects under these settings. We encourage sociologists to incorporate these insights into their empirical research.

Exploration of the factors that influence total phosphorus in surface water and an evaluation of surface water vulnerability based on an advanced algorithm and traditional index method

Due to the continuous influence of human activities, phosphorus pollution in surface water has become a persistent problem that needs to be addressed since phosphorous entails certain risks and degrees of damage to ecosystems and humans. The presence and accumulation of total phosphorus (TP) concentrations in surface waters is the result of a combined effect of many natural and anthropogenic factors, and it is often difficult to intuitively identify the individual importance of each factor in regard to the pollution of the aquatic environment. Considering these issues, this study provides a new methodology to better understand the vulnerability of surface water to TP pollution and the factors that influence TP pollution through the application of two modeling approaches. This includes the boosted regression tree (BRT), an advanced machine learning method, and the traditional comprehensive index method (CIM). Different factors, such as natural variables (including slope, soil texture, normalized difference vegetation index (NDVI), precipitation, and drainage density) and point and nonpoint source anthropogenic factors were included to model the vulnerability of surface water to TP pollution. Two methods were used to produce a vulnerability map of surface water to TP pollution. Pearson correlation analysis was used to validate the two vulnerability assessment methods. The results showed that BRT was more strongly correlated than CIM. In addition, the importance ranking results showed that slope, precipitation, NDVI, decentralized livestock farming and soil texture had a greater influence on TP pollution. Industrial activities, scale livestock farming and population density, which are all contributing sources of pollution, were all relatively less important. The introduced methodology can be used to quickly identify the area most vulnerable to TP pollution and to develop problem specific adaptive policies and measures to reduce the damage from TP pollution.

Best (but oft forgotten) statistical practices: Measuring real-world intervention effectiveness using electronic health data

The evidence base supporting the use of most interventions consists primarily of data from randomized controlled trials (RCTs), but how and to whom interventions are delivered in clinical practice may differ substantially from these foundational RCTs. With the increasing availability of electronic health data, it is now feasible to study the "real-world" effectiveness of a wide range of interventions. However, real-world intervention effectiveness studies using electronic health data face many challenges including data quality, selection bias, confounding by indication, and lack of generalizability. In this article, we describe the key barriers to generating high-quality evidence from real-world intervention effectiveness studies and suggest statistical best practices for addressing them.

Design of a five-layers multi-energy X-ray imaging detector for material sorting

X-ray transmission imaging (XRT) is widely used for sorting materials. However, conventional single-energy and dual-energy X-ray systems have poor ability to discriminate between materials with similar atomic number (Z), and the count rate of available multi-energy XRT detectors could not support high-speed industrial applications. This paper presents the design of a detector that can potentiality achieve high-speed multi-energy X-ray imaging using Geant4 simulations. This detector consisted of five detection layers (with three scintillator materials: CsI, GOS and CdWO₄), two metal filters, which allows X-ray imaging at five energies. Validation simulation showed that the 15% more accurate than a dual-layers detector in the classification of Mg and Al alloys.

Earnings Trajectories After Divorce: The Legacies of the Earner Model During Marriage

Divorce marks the legal endpoint of a marital union. While divorce is increasingly seen as a ‘clean break’, the past marital history of the couple may nevertheless shape their present conditions. In particular, there may be a legacy of a highly gendered division of labour during marriage that may affect the ex-spouses’ earning trajectories beyond the date of divorce. Using register data from the German Pension Fund, we examine the earning trajectories of heterosexual couples who filed for a divorce in 2013 (24,616 men and 24,616 women). Using fixed-effects and matching techniques, we compare the earning trajectories of divorcees with those of a control group of married persons in the period spanning two years before and two years after divorce. In particular, we examine how the earner models divorcees followed during marriage shaped their future earning trajectories. Our results show that, on average, the earnings of a divorced woman in a male breadwinner constellation increased after divorce, while the earnings of her male ex-spouse declined. Nevertheless, large gender differences in earnings persisted: 2 years after separation, a divorced woman who had been in a male breadwinner constellation was, on average, earning 72% less than her ex-spouse. We discuss our findings against the background of recent policy reforms in Germany, which assume that ex-partners should be economically ‘self-reliant’ after divorce.

Propensity score matching after multiple imputation when a confounder has missing data

One of the main challenges when using observational data for causal inference is the presence of confounding. A classic approach to account for confounding is the use of propensity score techniques that provide consistent estimators of the causal treatment effect under four common identifiability assumptions for causal effects, including that of no unmeasured confounding. Propensity score matching is a very popular approach which, in its simplest form, involves matching each treated patient to an untreated patient with a similar estimated propensity score, that is, probability of receiving the treatment. The treatment effect can then be estimated by comparing treated and untreated patients within the matched dataset. When missing data arises, a popular approach is to apply multiple imputation to handle the missingness. The combination of propensity score matching and multiple imputation is increasingly applied in practice. However, in this article we demonstrate that combining multiple imputation and propensity score matching can lead to over-coverage of the confidence interval for the treatment effect estimate. We explore the cause of this over-coverage and we evaluate, in this context, the performance of a correction to Rubin's rules for multiple imputation proposed by finding that this correction removes the over-coverage.

A latent class model to multiply impute missing treatment indicators in observational studies when inferences of the treatment effect are made using propensity score matching

Analysts often estimate treatment effects in observational studies using propensity score matching techniques. When there are missing covariate values, analysts can multiply impute the missing data to create m completed data sets. Analysts can then estimate propensity scores on each of the completed data sets, and use these to estimate treatment effects. However, there has been relatively little attention on developing imputation models to deal with the additional problem of missing treatment indicators, perhaps due to the consequences of generating implausible imputations. However, simply ignoring the missing treatment values, akin to a complete case analysis, could also lead to problems when estimating treatment effects. We propose a latent class model to multiply impute missing treatment indicators. We illustrate its performance through simulations and with data taken from a study on determinants of children's cognitive development. This approach is seen to obtain treatment effect estimates closer to the true treatment effect than when employing conventional imputation procedures as well as compared to a complete case analysis.

Propensity score analysis with missing data using a multi-task neural network

BACKGROUND

Propensity score analysis is increasingly used to control for confounding factors in observational studies. Unfortunately, unavoidable missing values make estimating propensity scores extremely challenging. We propose a new method for estimating propensity scores in data with missing values.

MATERIALS AND METHODS

Both simulated and real-world datasets are used in our experiments. The simulated datasets were constructed under 2 scenarios, the presence (T = 1) and the absence (T = 0) of the true effect. The real-world dataset comes from LaLonde's employment training program. We construct missing data with varying degrees of missing rates under three missing mechanisms: MAR, MCAR, and MNAR. Then we compare MTNN with 2 other traditional methods in different scenarios. The experiments in each scenario were repeated 20,000 times. Our code is publicly available at https://github.com/ljwa2323/MTNN .

RESULTS

Under the three missing mechanisms of MAR, MCAR and MNAR, the RMSE between the effect and the true effect estimated by our proposed method is the smallest in simulations and in real-world data. Furthermore, the standard deviation of the effect estimated by our method is the smallest. In situations where the missing rate is low, the estimation of our method is more accurate.

CONCLUSIONS

MTNN can perform propensity score estimation and missing value filling at the same time through shared hidden layers and joint learning, which solves the dilemma of traditional methods and is very suitable for estimating true effects in samples with missing values. The method is expected to be broadly generalized and applied to real-world observational studies.

Knowledge Discovery: Methods from data mining and machine learning

The interdisciplinary field of knowledge discovery and data mining emerged from a necessity of big data requiring new analytical methods beyond the traditional statistical approaches to discover new knowledge from the data mine. This emergent approach is a dialectic research process that is both deductive and inductive. The data mining approach automatically or semi-automatically considers a larger number of joint, interactive, and independent predictors to address causal heterogeneity and improve prediction. Instead of challenging the conventional model-building approach, it plays an important complementary role in improving model goodness of fit, revealing valid and significant hidden patterns in data, identifying nonlinear and non-additive effects, providing insights into data developments, methods, and theory, and enriching scientific discovery. Machine learning builds models and algorithms by learning and improving from data when the explicit model structure is unclear and algorithms with good performance are difficult to attain. The most recent development is to incorporate this new paradigm of predictive modeling with the classical approach of parameter estimation regressions to produce improved models that combine explanation and prediction.

Estimation of the average treatment effect with variable selection and measurement error simultaneously addressed for potential confounders

In the framework of causal inference, the inverse probability weighting estimation method and its variants have been commonly employed to estimate the average treatment effect. Such methods, however, are challenged by the presence of irrelevant pre-treatment variables and measurement error. Ignoring these features and naively applying the usual inverse probability weighting estimation procedures may typically yield biased inference results. In this article, we develop an inference method for estimating the average treatment effect with those features taken into account. We establish theoretical properties for the resulting estimator and carry out numerical studies to assess the finite sample performance of the proposed estimator.

CART model to classify the drought status of diverse tomato genotypes by VPD, air temperature, and leaf-air temperature difference

Regular water management is crucial for the cultivation of tomato (Solanum lycopersicum L.). Inadequate irrigation leads to water stress and a reduction in tomato yield and quality. Therefore, it is important to develop an efficient classification method of the drought status of tomato for the timely application of irrigation. In this study, a simple classification and regression tree (CART) model that includes air temperature, vapor pressure deficit, and leaf-air temperature difference was established to classify the drought status of three tomato genotypes (i.e., cherry type 'Tainan ASVEG No. 19', large fruits breeding line '108290', and wild accession 'LA2093'). The results indicate that the proposed CART model exhibited a higher predictive sensitivity, specificity, geometric mean, and accuracy performance compared to the logistic model. In addition, the CART model was applicable not only to three tomato genotypes but across vegetative and reproductive stages. Furthermore, while the drought status was divided into low, medium, and high, the CART model provided a higher predictive performance than that of the logistic model. The results suggest that the drought status of tomato can be accurately classified by the proposed CART model. These results will provide a useful tool of the regular water management for tomato cultivation.

Heterogeneous treatment effects analysis for social scientists: A review

Social scientists have long been interested in the varying responses to a specific intervention, motivating the enterprise of heterogeneous treatment effects (HTE) analysis. Over the past five decades, the rapid development of HTE methods, from conventional multiplicative interactions in linear models to explorations based on machine learning techniques, has been witnessed. This article presents a systematic review of major HTE methods, including multiplicative interaction modeling, generalized additive modeling, propensity-score-based methods, marginal treatment effect, separate LASSO constraints, causal trees, causal forests, Bayesian additive regression trees, and meta-learners (i.e., the S-learner, T-learner, X-learner, and R-learner). These methods, as described roughly in a chronological order to emphasize methodological developments, are addressed to highlight their respective strengths and limitations. Following an illustrative example, this article reflects on future methodological developments.

SMART choice (knee) tool: a patient-focused predictive model to predict improvement in health-related quality of life after total knee arthroplasty

BACKGROUND

Current predictive tools for TKA focus on clinicians rather than patients as the intended user. The purpose of this study was to develop a patient-focused model to predict health-related quality of life outcomes at 1-year post-TKA.

METHODS

Patients who underwent primary TKA for osteoarthritis from a tertiary institutional registry after January 2006 were analysed. The primary outcome was improvement after TKA defined by the minimal clinically important difference in utility score at 1-year post-surgery. Potential predictors included demographic information, comorbidities, lifestyle factors, and patient-reported outcome measures. Four models were developed, including both conventional statistics and machine learning (artificial intelligence) methods: logistic regression, classification tree, extreme gradient boosted trees, and random forest models. Models were evaluated using discrimination and calibration metrics.

RESULTS

A total of 3755 patients were included in the study. The logistic regression model performed the best with respect to both discrimination (AUC = 0.712) and calibration (intercept = -0.083, slope = 1.123, Brier score = 0.202). Less than 2% (n = 52) of the data were missing and therefore removed for complete case analysis. The final model used age (categorical), sex, baseline utility score, and baseline Veterans-RAND 12 responses as predictors.

CONCLUSION

The logistic regression model performed better than machine learning algorithms with respect to AUC and calibration plot. The logistic regression model was well calibrated enough to stratify patients into risk deciles based on their likelihood of improvement after surgery. Further research is required to evaluate the performance of predictive tools through pragmatic clinical trials.

LEVEL OF EVIDENCE

Level II, decision analysis.

Targeting resources efficiently and justifiably by combining causal machine learning and theory

Introduction

Efficient allocation of limited resources relies on accurate estimates of potential incremental benefits for each candidate. These heterogeneous treatment effects (HTE) can be estimated with properly specified theory-driven models and observational data that contain all confounders. Using causal machine learning to estimate HTE from big data offers higher benefits with limited resources by identifying additional heterogeneity dimensions and fitting arbitrary functional forms and interactions, but decisions based on black-box models are not justifiable.

Methods

Our solution is designed to increase resource allocation efficiency, enhance the understanding of the treatment effects, and increase the acceptance of the resulting decisions with a rationale that is in line with existing theory. The case study identifies the right individuals to incentivize for increasing their physical activity to maximize the population's health benefits due to reduced diabetes and heart disease prevalence. We leverage large-scale data from multi-wave nationally representative health surveys and theory from the published global meta-analysis results. We train causal machine learning ensembles, extract the heterogeneity dimensions of the treatment effect, sign, and monotonicity of its moderators with explainable AI, and incorporate them into the theory-driven model with our generalized linear model with the qualitative constraint (GLM_QC) method.

Results

The results show that the proposed methodology improves the expected health benefits for diabetes by 11% and for heart disease by 9% compared to the traditional approach of using the model specification from the literature and estimating the model with large-scale data. Qualitative constraints not only prevent counter-intuitive effects but also improve achieved benefits by regularizing the model.

Impact of enterprise digitalization on green innovation performance under the perspective of production and operation

Introduction

How enterprises should practice digitalization transformation to effectively improve green innovation performance is related to the sustainable development of enterprises and the economy, which is an important issue that needs to be clarified.

Methods

This research uses the perspective of production and operation to deconstruct the digitalization of industrial listed enterprises from 2016 to 2020 into six features. A variety of machine learning methods are used, including DBSCAN, CART and other algorithms, to specifically explore the complex impact of enterprise digitalization feature configuration on green innovation performance.

Conclusions

(1) The more advanced digitalization transformation the enterprises have, the more possibly the high green innovation performance can be achieved. (2) Digitalization innovation is the digitalization element with the strongest influence ability on green innovation performance. (3) As the advancement of digitalization transformation, enterprises should also focus on digitalization innovation input and digitalization operation output, otherwise they should pay attention to digitalization management and digitalization operation output.

Discussion

The conclusions of this research will help enterprises understand their digitalization competitiveness and how to practice digitalization transformation to enhance green innovation performance, and also help the government to formulate policies to promote the development of green innovation in the digital economy era.

Theory and practice of propensity score analysis

Propensity score analysis has been widely used in observational studies to make a causal inference. This study introduces three assumptions for causal inferences-conditional exchangeability, positivity, and consistency-and five steps for propensity score (PS) analysis-1) construct appropriate PS models, 2) check overlap in PS, 3) apply appropriate weighting (inverse probability of treatment weighting, standardized mortality ratio weighting, matching weights, and overlap weights) or matching methods according to the target of inference, 4) check the balance of covariates, and 5) estimate the effect of exposure appropriately. Finally, the advantages of PS analyses over conventional multivariable regression are discussed.

Estimation for volunteer web survey samples using a model-averaging approach

Propensity score approach is a popular technique for estimating the population based on volunteer web survey samples. Various models have been used to estimate propensity scores and produce different population estimates. To obtain more accurate population estimators, we propose a model-averaging estimation approach based on propensity score estimates from a parametric logistic regression model and a nonparametric generalized boosted model. Consistency and asymptotic normality of the proposed estimators are established. A computation algorithm is also developed to implement the proposed method. Simulation studies are conducted to compare the performance of the proposed method with the other methods. A survey data from the Netizen Social Awareness Survey (NSAS) is used to illustrate the proposed methodology.

Metabolite, protein, and tissue dysfunction associated with COVID-19 disease severity

Proteins are direct products of the genome and metabolites are functional products of interactions between the host and other factors such as environment, disease state, clinical information, etc. Omics data, including proteins and metabolites, are useful in characterizing biological processes underlying COVID-19 along with patient data and clinical information, yet few methods are available to effectively analyze such diverse and unstructured data. Using an integrated approach that combines proteomics and metabolomics data, we investigated the changes in metabolites and proteins in relation to patient characteristics (e.g., age, gender, and health outcome) and clinical information (e.g., metabolic panel and complete blood count test results). We found significant enrichment of biological indicators of lung, liver, and gastrointestinal dysfunction associated with disease severity using publicly available metabolite and protein profiles. Our analyses specifically identified enriched proteins that play a critical role in responses to injury or infection within these anatomical sites, but may contribute to excessive systemic inflammation within the context of COVID-19. Furthermore, we have used this information in conjunction with machine learning algorithms to predict the health status of patients presenting symptoms of COVID-19. This work provides a roadmap for understanding the biochemical pathways and molecular mechanisms that drive disease severity, progression, and treatment of COVID-19.

Machine Learning Methods in Health Economics and Outcomes Research-The PALISADE Checklist: A Good Practices Report of an ISPOR Task Force

Advances in machine learning (ML) and artificial intelligence offer tremendous potential benefits to patients. Predictive analytics using ML are already widely used in healthcare operations and care delivery, but how can ML be used for health economics and outcomes research (HEOR)? To answer this question, ISPOR established an emerging good practices task force for the application of ML in HEOR. The task force identified 5 methodological areas where ML could enhance HEOR: (1) cohort selection, identifying samples with greater specificity with respect to inclusion criteria; (2) identification of independent predictors and covariates of health outcomes; (3) predictive analytics of health outcomes, including those that are high cost or life threatening; (4) causal inference through methods, such as targeted maximum likelihood estimation or double-debiased estimation-helping to produce reliable evidence more quickly; and (5) application of ML to the development of economic models to reduce structural, parameter, and sampling uncertainty in cost-effectiveness analysis. Overall, ML facilitates HEOR through the meaningful and efficient analysis of big data. Nevertheless, a lack of transparency on how ML methods deliver solutions to feature selection and predictive analytics, especially in unsupervised circumstances, increases risk to providers and other decision makers in using ML results. To examine whether ML offers a useful and transparent solution to healthcare analytics, the task force developed the PALISADE Checklist. It is a guide for balancing the many potential applications of ML with the need for transparency in methods development and findings.

Dynamic Ensemble Selection for Imbalanced Data Streams With Concept Drift

Ensemble learning, as a popular method to tackle concept drift in data stream, forms a combination of base classifiers according to their global performances. However, concept drift generally occurs in local data space, causing significantly different performances of a base classifier at different locations. Thus, employing global performance as a criterion to select base classifier is inappropriate. Moreover, data stream is often accompanied by class imbalance problem, which affects the classification accuracy of ensemble learning on minority instances. To drawback these problems, a dynamic ensemble selection for imbalanced data streams with concept drift (DES-ICD) is proposed. For data arrived in chunk-by-chunk, a novel synthetic minority oversampling technique with adaptive nearest neighbors (AnnSMOTE) is developed to generate new minority instances that conform to the new concept. Following that, DES-ICD creates a base classifier on newly arrived data chunk balanced by AnnSMOTE and merges it with historical base classifiers to form a candidate classifier pool. For each query instance, the optimal combination is constructed in terms of the performance of candidate classifiers in its neighborhood. Experimental results for nine synthetic and five real-world datasets show that the proposed method outperforms seven comparative methods on classification accuracy and tracks new concepts in an imbalanced data stream more preciously.

Machine learning for improving high-dimensional proxy confounder adjustment in healthcare database studies: an overview of the current literature

Controlling for large numbers of variables that collectively serve as 'proxies' for unmeasured factors can often improve confounding control in pharmacoepidemiologic studies utilizing administrative healthcare databases. There is a growing body of evidence showing that data-driven machine learning algorithms for high-dimensional proxy confounder adjustment can supplement investigator-specified variables to improve confounding control compared to adjustment based on investigator-specified variables alone. Consequently, there has been a recent focus on the development of data-driven methods for high-dimensional proxy confounder adjustment. In this paper, we discuss the considerations underpinning three areas for data-driven high-dimensional proxy confounder adjustment: 1) feature generation-transforming raw data into covariates (or features) to be used for proxy adjustment; 2) covariate prioritization, selection and adjustment; and 3) diagnostic assessment. We survey current approaches and recent advancements within each area, including the most widely used approach to proxy confounder adjustment in healthcare database studies (the high-dimensional propensity score or hdPS). We also discuss limitations of the hdPS and outline recent advancements that incorporate the principles of proxy adjustment with machine learning extensions to improve performance. We further discuss challenges and avenues of future development within each area. This manuscript is endorsed by the International Society for Pharmacoepidemiology (ISPE). This article is protected by copyright. All rights reserved.