Multiple Treatment Effect Estimation using Deep Generative Model with Task Embedding

This paper addresses causal inference and modeling over event streams where data have high throughput, are unbounded, and may arrive out of order. The availability of large amount of data with these characteristics presents several new challenges related to causal modeling, such as the need for fast causal inference operations while ensuring consistent and valid results. There is no existing work specifically for such a streaming environment. We meet the challenges by introducing a time-centric causal inference strategy which leverages temporal precedence information to decrease the number of conditional independence tests required to establish the causalities between variables in a causal network. (Dependency and temporal precedence of cause over effect are the two properties of a causal relationship.) Moreover, we employ change-driven causal network inference to safely reduce the running time further. In this paper we present the Order-Aware Temporal Network Inference algorithm to model the temporal precedence relationships into a temporal network and then propose the Enhanced Fast Causal Network Inference algorithm for learning a causal network faster using the temporal network. Experiments using synthetic and real datasets demonstrate the efficacy of the proposed algorithms.

Machine learning algorithms can produce biased outcome/prediction, typically, against minorities and under-represented sub-populations. Therefore, fairness is emerging as an important requirement for the safe application of machine learning based technologies. The most commonly used fairness notions (e.g. statistical parity, equalized odds, predictive parity, etc.) are observational and rely on mere correlation between variables. These notions fail to identify bias in case of statistical anomalies such as Simpson's or Berkson's paradoxes. Causality-based fairness notions (e.g. counterfactual fairness, no-proxy discrimination, etc.) are immune to such anomalies and hence more reliable to assess fairness. The problem of causality-based fairness notions, however, is that they are defined in terms of quantities (e.g. causal, counterfactual, and path-specific effects) that are not always measurable. This is known as the identifiability problem and is the topic of a large body of work in the causal inference literature. The first contribution of this paper is a compilation of the major identifiability results which are of particular relevance for machine learning fairness. To the best of our knowledge, no previous work in the field of ML fairness or causal inference provides such systemization of knowledge. The second contribution is more general and addresses the main problem of using causality in machine learning, that is, how to extract causal knowledge from observational data in real scenarios. This paper shows how this can be achieved using identifiability.

A good deal of humility: Cochran on observational studies

The National Research Council’s report on the prevention and treatment of missing data highlighted the need to clearly specify causal estimands. This focus fundamentally changed how the missing data problem was perceived and addressed in clinical trials. The recent ICH E9(R1) addendum is another major step in promoting the use of the causal estimands framework that should further influence how clinical trial protocols and statistical analysis plans are written and implemented. The language of potential outcomes that is widely accepted in the causal inference literature is not widely recognized in the clinical trialists community and was not used in defining causal estimands in the NRC report or the ICH E9(R1). In this article, we attempt to bridge the gap between the causal inference community and clinical trialists to further advance the use of causal estimands in clinical trial settings. We illustrate how concepts from causal literature, such as potential outcomes and dynamic treatment regimens, can facilitate defining and implementing causal estimands and may provide a unifying language to describing the targets for both observational and randomized clinical trials.

This study compares the effects of single dose and multiple dose treatment of cyclophosphamide (CP) on oxidative stress-mediated follicular damage in mouse ovary. In the first experiment, adult female mice were administered with a single dose of CP (100 mg/kg body weight/mouse) and autopsied 72 hr after treatment. In the second experiment, adult female mice were injected with multiple doses of CP (40 mg/kg body weight/day/mouse for 10 consecutive days) and sacrificed on Day 11. There was a 58, 48, 53, and 51% loss of primordial, primary, preantral, and antral follicles, respectively, following the administration of a single dose of CP, whereas, multiple dose of CP caused only 35% reduction in primordial follicles coupled with 28, 23, and 38%, loss of primary, preantral, and antral follicles, respectively. There was a decrease in activities of the ovarian antioxidant enzymes and increase in reactive oxygen species (ROS) and malondialdehyde (MDA) concentrations following single dose CP, whereas multiple dose treatment caused an increase in activities of these enzymes and decrease in ROS and MDA concentrations. The serum concentration of estradiol was significantly decreased following single or multiple dose treatment. The ovarian damage caused by a single high dose of CP administration is higher than that by multiple doses of smaller amount, though the total amount of CP administered was higher with multiple treatment. The results of the current study reveal the benefit of metronomic chemotherapy in cancer treatment, for its effectiveness in reducing ovarian toxicity, a major side effect in young female patients.

Automatic image segmentation and quantification are critical steps in medical image analysis. The main challenges in medical image segmentation are due to the imbalance in data distribution and spatial variations of ROI. The ideal segmentation should extract all kinds of segments irrespective of size, shape and position. Commonly used metrics such as accuracy, IOU, Dice similarity coefficient consider all the detected pixels in a similar way. However, the detection of smaller segments is critical in medical analysis since it helps in early treatment of the disease and are also easier to miss. Hence, segmentation evaluation must accord larger weighting to pixels in smaller segments compared to the bigger ones. We propose a novel evaluation metric for segmentation performance, emphasizing smaller segments, by assigning a higher weightage to those pixels. Weighted false positives are also considered in deriving the new metric named, “SSEGEP” (Smatt SEGment Emphasized Performance evaluation metric), (range: 0 (Bad) to 1 (Good)). The proposed approach has been applied to two different publicly available real medical data sets of CT modality consisting of scans of the liver and pancreas of 131 and 107 subjects respectively and the results have been compared with existing evaluation metrics. Statistical significance testing is performed to quantity the relevance of the proposed approach. In comparison to Dice similarity coefficient, SSEGEP resulted in a promising p-value of the order 10-18 for hepatic tumor. The proposed metric is found to perform better for the images having multiple segments for a single label and where the regions of interest are not localized.

Time series data are a collection of chronological observations which are generated by several domains such as medical and financial fields. Over the years, different tasks such as classification, forecasting and clustering have been proposed to analyze this type of data. Time series data have been also used to study the effect of interventions overtime. Moreover, in many fields of science, learning the causal structure of dynamic systems and time series data is considered an interesting task which plays an important role in scientific discoveries. Estimating the effect of an intervention and identifying the causal relations from the data can be performed via causal inference. Existing surveys on time series discuss traditional tasks such as classification and forecasting or explain the details of the approaches proposed to solve a specific task. In this paper, we focus on two causal inference tasks, i.e., treatment effect estimation and causal discovery for time series data and provide a comprehensive review of the approaches in each task. Furthermore, we curate a list of commonly used evaluation metrics and datasets for each task and provide an in-depth insight. These metrics and datasets can serve as benchmark for research in the field.

Based on benchmarking various methods and analyzing multiple real scRNA-seq datasets, a computational platform/workflow and a set of tips for best practices are developed for analyzing causal interactions or relationships in single cells.

Psychologists leverage longitudinal designs to examine the causal effects of a focal predictor (i.e., treatment or exposure) over time. But causal inference of naturally observed time-varying treatments is complicated by treatment-dependent confounding in which earlier treatments affect confounders of later treatments. In this tutorial article, we introduce psychologists to an established solution to this problem from the causal inference literature: the parametric g-computation formula. We explain why the g-formula is effective at handling treatment-dependent confounding. We demonstrate that the parametric g-formula is conceptually intuitive, easy to implement, and well-suited for psychological research. We first clarify that the parametric g-formula essentially utilizes a series of statistical models to estimate the joint distribution of all post-treatment variables. These statistical models can be readily specified as standard multiple linear regression functions. We leverage this insight to implement the parametric g-formula using lavaan, a widely adopted R package for structural equation modeling. Moreover, we describe how the parametric g-formula may be used to estimate a marginal structural model whose causal parameters parsimoniously encode time-varying treatment effects. We hope this accessible introduction to the parametric g-formula will equip psychologists with an analytic tool to address their causal inquiries using longitudinal data.

Advertising effectiveness measurement is a fundamental problem in online advertising. Various causal inference methods have been employed to measure the causal effects of ad treatments. However, existing methods mainly focus on linear logistic regression for univariate and binary treatments and are not well suited for complex ad treatments of multi-dimensions, where each dimension could be discrete or continuous. In this paper we propose a novel two-stage causal inference framework for assessing the impact of complex ad treatments. In the first stage, we estimate the propensity parameter via a sparse additive model; in the second stage, a propensity-adjusted regression model is applied for measuring the treatment effect. Our approach is shown to provide an unbiased estimation of the ad effectiveness under regularity conditions. To demonstrate the efficacy of our approach, we apply it to a real online advertising campaign to evaluate the impact of three ad treatments: ad frequency, ad channel, and ad size. We show that the ad frequency usually has a treatment effect cap when ads are showing on mobile device. In addition, the strategies for choosing best ad size are completely different for mobile ads and online ads.

Ecology has yet to embrace causal inference, yet most questions in ecology are causal. Despite the common use of terms that imply causation, such as "shapes", "drives", or "impacts", many studies shy away from directly acknowledging their causal ambitions. This avoidance not only obscures the true intent of research but also underpins a broader challenge within the field's approach to science. Ecology relies heavily on observational data, and so the necessity for robust causal inference becomes paramount. However, causal methods are also needed for non-randomised experiments. We critique the predominance in ecology of scientifically empty statistical procedures that lack scientific clarity and value. We advocate for a shift towards explicit causal inference, arguing that understanding causality is not confined to randomised controlled trials but can also be enriched through observational data when paired with rigorous causal inference methodologies. This paper elucidates the common pitfalls in ecological studies, such as throwing all variables into an analysis, use of the Akaike information criterion (AIC) for model selection, the "Table 2 fallacy" and the misuse of controls: all of which can lead to misleading scientific understanding. The good news is that causal inference is not primarily a statistical problem, but rather a scientific one that is accessible to all ecologists. We can achieve reasonable progress by continuing to use the standard statistical toolbox based around regression models, familiar to many ecologists, paired with causal diagrams. For regression, causal inference is about understanding what we should condition on (good controls) and what we should not condition on (bad controls). We provide not only a critique but a constructive guide, aiming to demystify causal inference and encourage its adoption in ecological studies using familiar approaches. By doing so, we seek to elevate the quality and impact of ecological research, moving beyond routine convenient statistical procedures and towards a more scientifically sound and insightful understanding of ecology.

Causal inference aims to estimate the causal relationships and effect sizes among treatments and outcomes. Electronic health record (EHR) data is a valuable healthcare data source that can support causal inference. However, a large percentage of the data is missing in EHRs and they are missing not at random (MNAR). Ignoring MNAR can lead to severe biases, to the extent where the causal structure underlying the data gets distorted. We proposed a new causal inference methodology that addresses the MNAR problem and thus helps preserve the causal structure. We compared the performance of our proposed method with the traditional causal inference method, structural equation modeling (SEM). We evaluated these methods for their accuracy in estimating the causal effect sizes and their ability to converge at all. We employed both simulation studies and real-world EHR data sets. We demonstrated that imputation under the improper missingness mechanism distorted the causal structure to a degree where SEM found it incompatible with the data and failed to. converge. Even when 20 to 30 % of the values were missing, SEM failed to converge in as many as 50% of the runs. The proposed causal inference method achieved a higher convergence rate and more accurate estimation of latent treatment effects both on the synthetic data and on a real EHR data set. We proposed a new methodology that incorporates the knowledge of missing data mechanisms. It significantly mitigated the biases associated with MNAR in the EHR dataset and substantially outperformed SEM that uses the improper missing data mechanism.

Previous chapter Next chapter Full AccessProceedings Proceedings of the 2022 SIAM International Conference on Data Mining (SDM)Learning Infomax and Domain-Independent Representations for Causal Effect Inference with Real-World DataZhixuan Chu, Stephen L. Rathbun, and Sheng LiZhixuan Chu, Stephen L. Rathbun, and Sheng Lipp.433 - 441Chapter DOI:https://doi.org/10.1137/1.9781611977172.49PDFBibTexSections ToolsAdd to favoritesExport CitationTrack CitationsEmail SectionsAboutAbstract The foremost challenge to causal inference with real-world data is to handle the imbalance in the covariates with respect to different treatment options, caused by treatment selection bias. To address this issue, recent literature has explored domain-invariant representation learning based on different domain divergence metrics (e.g., Wasserstein distance, maximum mean discrepancy, position-dependent metric, and domain overlap). In this paper, we reveal the weaknesses of these strategies, i.e., they lead to the loss of predictive information when enforcing the domain invariance; and the treatment effect estimation performance is unstable, which heavily relies on the characteristics of the domain distributions and the choice of domain divergence metrics. Motivated by information theory, we propose to learn the Infomax and Domain-Independent Representations to solve the above puzzles. Our method utilizes the mutual information between the global feature representations and individual feature representations, and the mutual information between feature representations and treatment assignment predictions, in order to maximally capture the common predictive information for both treatment and control groups. Moreover, our method filters out the influence of instrumental and irrelevant variables, and thus it effectively increases the predictive ability of potential outcomes. Experimental results on both the synthetic and real-world datasets show that our method achieves state-of-the-art performance on causal effect inference. Moreover, our method exhibits reliable prediction performances when facing data with different characteristics of data distributions, complicated variable types, and severe covariate imbalance. Previous chapter Next chapter RelatedDetails Published:2022eISBN:978-1-61197-717-2 https://doi.org/10.1137/1.9781611977172Book Series Name:ProceedingsBook Code:PRDT22Book Pages:1-737Key words:causal inference, selection bias, mutual information, real-world data

Mediation Analysis with Multiple Mediators.

ABSTRACTFrequentists’ inference often delivers point estimators associated with confidence intervals or sets for parameters of interest. Constructing the confidence intervals or sets requires understanding the sampling distributions of the point estimators, which, in many but not all cases, are related to asymptotic Normal distributions ensured by central limit theorems. Although previous literature has established various forms of central limit theorems for statistical inference in super population models, we still need general and convenient forms of central limit theorems for some randomization-based causal analyses of experimental data, where the parameters of interests are functions of a finite population and randomness comes solely from the treatment assignment. We use central limit theorems for sample surveys and rank statistics to establish general forms of the finite population central limit theorems that are particularly useful for proving asymptotic distributions of randomization tests under the sharp null hypothesis of zero individual causal effects, and for obtaining the asymptotic repeated sampling distributions of the causal effect estimators. The new central limit theorems hold for general experimental designs with multiple treatment levels, multiple treatment factors and vector outcomes, and are immediately applicable for studying the asymptotic properties of many methods in causal inference, including instrumental variable, regression adjustment, rerandomization, cluster-randomized experiments, and so on. Previously, the asymptotic properties of these problems are often based on heuristic arguments, which in fact rely on general forms of finite population central limit theorems that have not been established before. Our new theorems fill this gap by providing more solid theoretical foundation for asymptotic randomization-based causal inference. Supplementary materials for this article are available online.