Causal Methods Research Articles

Target Trial Emulation: Impact of C-Section on Birth Outcomes for Spontaneous Preterm Breech Presentations

Breech presentation is a risk factor for poor birth outcomes. Guidance for safest mode of delivery in breech presenting full term pregnancies is available but there is a lack of evidence on the optimal mode of delivery in breech presenting preterm spontaneous births. The aim of this study is to evaluate the causal effect of mode of delivery on extended perinatal death (EPD) in this population using a trial emulation. Randomized experiments are the gold standard for estimating causal effects; however, previous attempts to conduct them in this context has resulted in recruitment challenges. Causal inference methods provide a framework for drawing causal conclusions from observational data. A trial emulation aims to overcome limitations common to observational studies by specifying core components of a randomized trial using observational data. Our trial emulation uses national Scottish health data to identify an eligible study sample and adjust for confounding characteristics between groups undergoing caesarean and vaginal deliveries to simulate randomized assignment of mode of delivery. We were able to estimate the per-protocol impact of caesarean delivery on EPD. Our data, containing all recorded potentially eligible births in Scotland between 1997 and 2019, was provided by eDRIS in a trusted research environment. After applying our eligibility criteria, our data contained 3,457 birth records, (2,361 c-section and 1,096 vaginal deliveries). There were 212 EPDs (1.7% (39) of c-sections and 16% (173) of vaginal deliveries). Our preliminary results suggest that caesarean delivery reduces the odds of EPD by 73% among breech presenting spontaneous preterm births.

How to Assess Digitisation Impacts on the Food Industry: The Dutch Case of Microdata Collection for Firms’ Sustainability

SummaryDigital innovations can increase resilience and sustainability of food businesses due to for instance increasing efficiency, reducing pollution or improving transparency about production conditions. However, much of the research on this topic has been done at the primary production level, i.e. on technology impacts on farms, or remains at an ex‐ante case study level when looking at digital technology impacts in downstream food businesses. Therefore, ex‐post impacts of digital technologies after their adoption are largely unknown.We here argue that this knowledge gap arises from the fact that we lack single company data (i.e. micro‐data) on the adoption of digital technologies in food businesses. Only with this information can actual impacts of these technologies be assessed using causal inference methods. As a way forward, we highlight a unique dataset of digital technology adoption among Dutch downstream food businesses that is available for research purposes. We show that digital technologies already have a considerable market penetration in the Netherlands. Therefore, we urge the scientific community and governmental statistical offices to collect and make available similar data globally and make use of the already existing data. Such impact assessments will inform and accelerate the transition towards more sustainable and resilient food value chains.

Exploring maladaptive patterns of small-scale green roofs through evaluation in a capacity of heat mitigation: A case study in seoul

Climate change leads to more frequent and intense heat waves, exposing urban populations to extreme heat conditions and significant health risks. Many cities are adopting Nature-based Solutions (NbS) to mitigate urban heat, with green roofs emerging as a widely recognized NbS. Although numerous green roof projects are implemented on a small scale, research on the effectiveness of small-scale green roofs on an urban scale is limited, making it challenging to identify the critical factors for both maladaptation and success. Therefore, we assessed the cooling potential of 18 small-scale green roofs in Seoul, Republic of Korea. This comparative study that adopted a multi-site approach examined the cooling capacity of green roofs to reduce surface temperature and identified characteristics of effective and ineffective green roofs. We utilized a developed difference-in-differences method to improve causal inference, effectively isolating the effects of individual green roofs from background climate change. The multi-site comparative approach and more robust causal inference methods improved our understanding of the effects of small-scale green roofs. The findings of this study indicate that three out of 18 green roofs were statistically significant maladaptation cases with an increase in LST. This evidence can help an urban planner reduce ineffectiveness and enhance effective adaptation practices. Our proposed method is expected to support government projects, especially those with limited budgets, in efficiently managing urban heat and reducing trial and error.

Pengaruh Empathy Response, Anonymity, dan Customization terhadap Customer Trust pada Chatbot Customer Service Tokopedia

Artificial intelligence (AI) technology has grown rapidly in customer service, including AI chatbots. Tokopedia has adopted an AI chatbot for customer service, but there are still many users who do not trust it because it lacks reliability. This study aims to examine the effect of AI chatbot dimensions on customer trust in Tokopedia, focusing on the variables of empathy response, anonymity, customization, and interaction. This research is quantitative using descriptive and causal methods, involving 437 Tokopedia user respondents through structured questionnaires. Data analysis using the SEM-PLS model with SmartPLS. The results show that empathy response, anonymity, customization, and interaction have a positive and significant influence on customer trust. In addition, empathy response, anonymity, and customization also have a positive effect on interaction. The suggestion for Tokopedia is to improve the quality of chatbots in providing empathetic responses, such as greeting users, appreciating, and validating their problems. Innovation in the aspects of empathy response, anonymity, customization, and interaction is also important. Future research could explore additional factors that influence customer trust as well as the mediating or moderating effects of certain variables.

Efektivitas Penghapusan Denda Administrasi Piutang Pajak Bumi dan Bangunan di Kota Bandung

Artificial intelligence (AI) technology has grown rapidly in customer service, including AI chatbots. Tokopedia has adopted an AI chatbot for customer service, but there are still many users who do not trust it because it lacks reliability. This study aims to examine the effect of AI chatbot dimensions on customer trust in Tokopedia, focusing on the variables of empathy response, anonymity, customization, and interaction. This research is quantitative using descriptive and causal methods, involving 437 Tokopedia user respondents through structured questionnaires. Data analysis using the SEM-PLS model with SmartPLS. The results show that empathy response, anonymity, customization, and interaction have a positive and significant influence on customer trust. In addition, empathy response, anonymity, and customization also have a positive effect on interaction. The suggestion for Tokopedia is to improve the quality of chatbots in providing empathetic responses, such as greeting users, appreciating, and validating their problems. Innovation in the aspects of empathy response, anonymity, customization, and interaction is also important. Future research could explore additional factors that influence customer trust as well as the mediating or moderating effects of certain variables.

A hybrid constrained continuous optimization approach for optimal causal discovery from biological data.

Understanding causal effects is a fundamental goal of science and underpins our ability to make accurate predictions in unseen settings and conditions. While direct experimentation is the gold standard for measuring and validating causal effects, the field of causal graph theory offers a tantalizing alternative: extracting causal insights from observational data. Theoretical analysis has shown that this is indeed possible, given a large dataset and if certain conditions are met. However, biological datasets, frequently, do not meet such requirements but evaluation of causal discovery algorithms is typically performed on synthetic datasets, which they meet all requirements. Thus, real-life datasets are needed, in which the causal truth is reasonably known. In this work we first construct such a large-scale real-life dataset and then we perform on it a comprehensive benchmarking of various causal discovery methods. We find that the PC algorithm is particularly accurate at estimating causal structure, including the causal direction which is critical for biological applicability. However, PC does only produces cause-effect directionality, but not estimates of causal effects. We propose PC-NOTEARS (PCnt), a hybrid solution, which includes the PC output as an additional constraint inside the NOTEARS optimization. This approach combines PC algorithm's strengths in graph structure prediction with the NOTEARS continuous optimization to estimate causal effects accurately. PCnt achieved best aggregate performance across all structural and effect size metrics. https://github.com/zhu-yh1/PC-NOTEARS.

O19 CAUSAL INFERENCE FOR HYPERTENSION DETECTION WITH WEARABLE DEVICES

Background and Objective: Hypertension is a leading cause of cardiovascular disease and premature death worldwide and it puts a heavy burden on the healthcare system. It is, therefore, very important to detect and evaluate hypertension and related cardiovascular events so as for early prevention, detection and management. Hypertension can be evaluated in real time with wearable noninvasive cardiac signals, such as electro-cardiogram (ECG) and photoplethysmography (PPG). Most previous studies predicted hypertension from ECG and PPG signals with extracted features that are correlated with hypertension. However, correlation is sometimes unreliable and may be affected by confounding factors. In this study, we explored the feasibility of predicting the risk of hypertension using causal inference methods. Additionally, we paid special attention to and verified the reliability and effectiveness of causality compared to correlation. Methods: Firstly, we constructed causal graphs by the Greedy Equivalence Search algorithm, and then applied causal strategies to obtain the optimal causal graph. Finally, we used machine learning classification algorithms to achieve hypertension prediction. Results: The machine learning classification models achieve great classification performance, with accuracy being 0.89, precision being 0.92, recall being 0.82, and F1-score being 0.87, which outperformed the correlation-based hypertension detection. Conclusions: The results indicate that the causal inference-based approach can potentially clarify the mechanism of hypertension detection with noninvasive signal and effectively detect hypertension. In addition, the results also re-veal that causality is more reliable and effective compared to correlation.

EXPOSURE EFFECTS ON COUNT OUTCOMES WITH OBSERVATIONAL DATA, WITH APPLICATION TO INCARCERATED WOMEN.

Causal inference methods can be applied to estimate the effect of a point exposure or treatment on an outcome of interest using data from observational studies. For example, in the Women's Interagency HIV Study, it is of interest to understand the effects of incarceration on the number of sexual partners and the number of cigarettes smoked after incarceration. In settings like this where the outcome is a count, the estimand is often the causal mean ratio, i.e., the ratio of the counterfactual mean count under exposure to the counterfactual mean count under no exposure. This paper considers estimators of the causal mean ratio based on inverse probability of treatment weights, the parametric g-formula, and doubly robust estimation, each of which can account for overdispersion, zero-inflation, and heaping in the measured outcome. Methods are compared in simulations and are applied to data from the Women's Interagency HIV Study.

Pengaruh Influencer Media Sosial Instagram terhadap Niat Beli Konsumen Produk Scarlett yang Dimediasi oleh Kredibilitas

This study aims to explore the influence of Instagram social media influencers on consumer purchase intentions of Scarlett products, with influencer credibility as a mediating variable. A quantitative approach with descriptive and causal methods was used in this study. Data analysis was carried out using SmartPLS 4.0. The results showed that influencer credibility significantly mediates the relationship between the influence of social media influencers and consumer purchase intentions of Scarlett products. Conclusion, 1) Respondents rated the variables of Instagram social media influencers, influencer credibility, and purchase intentions of Scarlett products in the good category. 2) Instagram social media influencers have a positive and significant influence on consumer purchase intentions of Scarlett products. 3) Influencer credibility also has a positive and significant influence on purchase intentions. 4) Influencer credibility effectively mediates the positive influence of Instagram social media influencers on consumer purchase intentions of Scarlett products. Keywords: Social Media Influencers, Credibility, Purchase Intention

Transcriptomic Module Discovery of Diarrhea-Predominant Irritable Bowel Syndrome: A Causal Network Inference Approach.

Irritable bowel syndrome with diarrhea (IBS-D) is the most prevalent subtype of IBS, characterized by chronic gastrointestinal symptoms in the absence of identifiable pathological findings. This study aims to investigate the molecular mechanisms underlying IBS-D using transcriptomic data. By employing causal network inference methods, we identify key transcriptomic modules associated with IBS-D. Utilizing data from public databases and applying advanced computational techniques, we uncover potential biomarkers and therapeutic targets. Our analysis reveals significant molecular alterations that affect cellular functions, offering new insights into the complex pathophysiology of IBS-D. These findings enhance our understanding of the disease and may foster the development of more effective treatments.

Causal deconfounding deep reinforcement learning for mobile robot motion planning

Deep reinforcement learning (DRL) has emerged as an efficient approach for motion planning in mobile robot systems. It leverages the offline training process to enhance real-time computation efficiency. In DRL-based methods, the DRL models are trained to compute an action based on the current state of the robot and the surrounding obstacles. However, the trained models may capture spurious correlations through potential confounders, resulting in non-robust state representations, which limits the models’ robustness and generalizability. In this paper, we propose a Causal Deconfounding DRL method for Motion Planning, CD-DRL-MP, to address spurious correlations and learn robust and generalizable policies. Specifically, we formalize the temporal causal relationships between states and actions using a structural causal model. We then extract the minimal sufficient state representation set by blocking the backdoor paths in the causal model. Finally, using the representation set, CD-DRL-MP learns the causal effect between states and actions while mitigating the detrimental influence of potential confounders and computes motion commands for mobile robots. Comprehensive experiments show that the proposed method significantly outperforms non-causal DRL methods and existing causal methods, while guaranteeing good robustness and generalizability.

Efficacy of the Dapivirine Vaginal Ring Accounting for Imperfect Adherence.

Product adherence is critical to obtaining objective estimates of efficacy of pre-exposure prophylactic interventions against HIV-1 infection. With imperfect adherence, intention-to-treat analyses assess the collective effects of complete, sub-optimal and non-adherence, providing a biased and attenuated estimate of the averagecausal effect of an intervention. Using data from the MTN-020/ASPIRE phase III trial evaluating HIV-1 efficacy of the dapivirine vaginal ring, we conducted per-protocol, and adherence-adjusted causal inference analyses using principal stratification and marginal structural models. We constructed two adherence cut offs of ≥ 0.9mg (low cutoff) and > 4.0mg (high cutoff) that represent drug released from the ring over a 28-day period. The HIV-1 efficacy estimate (95% CI) was 30.8% (3.6%, 50.3%) (P = 0.03) from the per-protocol analysis, and 53.6% (16.5%, 74.3%) (P = 0.01) among the highest predicted adherers from principal stratification analyses using the low cutoff. Marginal structural models produced efficacy estimates (95% CIs)ranging from 48.8 (21.8, 66.4) (P = 0.0019) to 56.5% (32.8%, 71.9%) (P = 0.0002). Application of adherence-adjusted causal inference methods are useful in interpreting HIV-1 efficacy in secondary analyses of PrEP clinical trials.

Pengaruh Shortage Cargo terhadap Kegiatan Transhipment Batu Bara di PT. Kartika Samudra Adijaya Site Berau

One of the most common obstacles is the difference in the amount of coal transported during the transhipment process compared to before the transhipment was carried out. This difference tends to lead to a reduction in the amount of coal (shortage cargo) which is certainly very detrimental to the recipient company which causes the transhipment process to occur more than once, thus disrupting the company's schedule and operations, resulting in the parties involved experiencing financial losses. This study was conducted to determine the influence of draught survey factors, hydrostatic table and cargo losses and how the influence of each factor simultaneously causes cargo shortages in transhipment activities, testing using comparative causal quantitative research methods using multiple linear regression data analysis techniques which were then processed using the SPSS version 23 program. The results obtained indicate that there is an influence of draught survey, hydrostatic table and cargo losses on transhipment activities and partially have an influence of 9%, 1.6% and 2.5% respectively. The variables of draught survey, hydrostatic table and cargo losses simultaneously cause cargo shortages so that it affects transhipment by 71.2%. while the remaining 28.8% is suspected to be influenced by other factors outside the research variables.

Genetic Architecture of Neurological Disorders and Their Endophenotypes: Insights from Genetic Association Studies.

Population-scale genetic association studies of complex neurologic diseases have identified the underlying genetic architecture as multifactorial. Despite the study sample sizes reaching the millions, the identified disease-related genes explain only a small fraction of the phenotypic variance. Notable advancements in statistical methods now enable researchers to gain insights even from genomic regions where genotype-phenotype associations do not reach statistical significance. Such studies confirm a highly interconnected molecular network comprising a core group of genes directly involved in the disease process, alongside an expanded peripheral network, each contributing a small but potentially important (modulatory) effect. Additionally, causal inference methods, utilizing genetic instruments, have shed light on putative causal links between risk factors and clinical endpoints. In light of the pervasive genetic overlap or pleiotropy, however, caution is warranted in interpreting causal relationships inferred from these analyses. In this chapter, I will introduce the genetic association model, provide insights into the current state of genetic association studies, and discuss potential future directions.

The delayed and combinatorial response of online public opinion to the real world: An inquiry into news texts during the COVID-19 era

In sociological research based on online public opinion, scholars often overlook the delay and combinatory nature of online responses to real-world events. This study aims to explore the delayed and combinatory responses of online public opinion to the intensity of the COVID-19 pandemic. Specifically, we seek to answer the following questions: (a) Is there a temporal delay in the response of online public opinion to the intensity of the pandemic? (b) Does this delay exhibit general characteristics of social networks, such as combinatory effects and higher-order interactions? To address these questions, we employ natural language processing techniques to extract online public opinion data and utilize statistical and machine learning-based causal inference methods for analysis. The findings indicate that online public opinion’s response to the intensity of COVID-19 is not immediate but rather exhibits a long-term lag. Identical COVID-19 intensity data can trigger multiple delayed public opinion responses, while a single delayed public opinion datum may be influenced by multiple preceding COVID-19 intensity data points. This delayed response of online public opinion and its higher-order network characteristics result in a waveform structure of real-world impacts influenced by online public opinion. We also utilized machine learning causal inference techniques to investigate the sensitivity differences in online public opinion responses to COVID-19 during various time periods.

Codiscovering graphical structure and functional relationships within data: A Gaussian Process framework for connecting the dots

Most problems within and beyond the scientific domain can be framed into one of the following three levels of complexity of function approximation. Type 1: Approximate an unknown function given input/output data. Type 2: Consider a collection of variables and functions, some of which are unknown, indexed by the nodes and hyperedges of a hypergraph (a generalized graph where edges can connect more than two vertices). Given partial observations of the variables of the hypergraph (satisfying the functional dependencies imposed by its structure), approximate all the unobserved variables and unknown functions. Type 3: Expanding on Type 2, if the hypergraph structure itself is unknown, use partial observations of the variables of the hypergraph to discover its structure and approximate its unknown functions. These hypergraphs offer a natural platform for organizing, communicating, and processing computational knowledge. While most scientific problems can be framed as the data-driven discovery of unknown functions in a computational hypergraph whose structure is known (Type 2), many require the data-driven discovery of the structure (connectivity) of the hypergraph itself (Type 3). We introduce an interpretable Gaussian Process (GP) framework for such (Type 3) problems that does not require randomization of the data, access to or control over its sampling, or sparsity of the unknown functions in a known or learned basis. Its polynomial complexity, which contrasts sharply with the super-exponential complexity of causal inference methods, is enabled by the nonlinear ANOVA capabilities of GPs used as a sensing mechanism.

Effect of smoking on lung cancer: A causal inference approach

Smoking is associated with an increased chance of developing lung cancer. Three causal inference methods, backdoor adjustment, front-door adjustment, and counterfactual are used to analyze observational data on smoking, lung cancer, and related risks factors. Backdoor adjustment fails to allow for possible presence of unobserved confounders, which is merited by front-door adjustment. Counterfactual harnesses individual patient statistics to establish causal relationships between smoking and cancer on the individual level, so as to evaluate lung cancer risks after changes in individual smoking habits. Results by different methodology are in good agreement and showcase a strong causation between smoking and lung cancer at both group and individual level.

Target Trial Emulation: A Call for More Widespread Use.

Causal inference methods intended for use with observational data have been widely available for decades, but barriers exist to their widespread adoption. These likely include lack of familiarity with several methodologic techniques often used in combination in these investigations such as inverse probability of treatment weighting and g-estimation, and the intensity of computational effort to employ these techniques. Even with these methods, critical design flaws undermine the ability to make valid causal inference in some studies. Identification of the need to explicitly pair study design elements with these causal inference methods led to development of a methodologic approach recently termed target trial emulation. This approach requires that investigators define a hypothetical randomized trial, emulate that hypothetical protocol in assembling the cohort and defining study elements and then conduct an analysis that attempts to mirror random treatment assignment. In the article by Heindel et al. (Am J Epidemiol. 2024;XXX(XX):XXXX-XXXX), the authors successfully emulate a target trial of systemic heparin during arteriovenous fistula creation on short-term endpoints by utilizing two existing randomized trials with key confounders available. Target trial emulation provides a framework to promote valid inference and generate high-quality contributions to the literature and its use should be expanded.

PresAIse, a prescriptive AI solution for enterprise

Prescriptive artificial intelligence (AI) represents a transformative shift in decision-making by offering causal insights and actionable recommendations. Despite its huge potential, enterprise adoption of prescriptive AI faces several challenges. One such challenge is caused by the lack of experimental data for many enterprises, making it hard to attribute differences in outcomes to interventions alone. The second pertains to the explainability of AI recommendations, which is crucial for enterprise decision-making settings. The third challenge is contributed by the silos between technologists and business users, hindering effective collaboration. This paper outlines an initiative from IBM Research, PresAIse, aiming to address some of these challenges by offering a suite of prescriptive AI solutions. Leveraging insights from various research papers, the solutions include scalable causal inference methods, interpretable decision-making approaches, and the integration of large language models (LLMs) to bridge the communication gap via a conversation agent. A proof-of-concept demonstrates the solutions’ potential by enabling non-ML experts to interact with prescriptive AI models via a natural language interface, democratizing advanced analytics for strategic decision-making.

Robust inference for causal mediation analysis of recurrent event data.

Recurrent events, including cardiovascular events, are commonly observed in biomedical studies. Understanding the effects of various treatments on recurrent events and investigating the underlying mediation mechanisms by which treatments may reduce the frequency of recurrent events are crucial tasks for researchers. Although causal inference methods for recurrent event data have been proposed, they cannot be used to assess mediation. This study proposed a novel methodology of causal mediation analysis that accommodates recurrent outcomes of interest in a given individual. A formal definition of causal estimands (direct and indirect effects) within a counterfactual framework is given, and empirical expressions for these effects are identified. To estimate these effects, a semiparametric estimator with triple robustness against model misspecification was developed. The proposed methodology was demonstrated in a real-world application. The method was applied to measure the effects of two diabetes drugs on the recurrence of cardiovascular disease and to examine the mediating role of kidney function in this process.