Causal Model Research Articles (Page 1)

ABSTRACT Quantitative intelligence analysis often leans on pattern recognition; in adversary-shaped settings, such correlations can be engineered. Building on Judea Pearl’s structural causal models – originally developed for natural data-generating processes – this study makes identification, not estimation, the gate to credible claims. Across operational cases, it clarifies when effects are recoverable from observational data and when additional leverage is required. The framework extends to sequential and multi-agent settings – as developed by Elias Bareinboim and colleagues – in which strategies are modelled as interventions. The result is an explicit, testable mapping from action to outcome that turns uncertain signals into actionable intelligence and delivers decision advantage.

ABSTRACT The current study presents a causal model of climate‐induced psychological resilience by assessing the impact of dispositional mindfulness, self‐efficacy beliefs, and perceived restorativeness of nature. It further examines the associations among climate change coping strategies, subjective well‐being components, eco‐emotions, and resilience. Structured questionnaires were administered to 552 participants selected through stratified sampling in a cross‐sectional design. The hypothesized model was tested through Structural Equation Modeling, with additional analyses addressing common method bias and evaluating construct reliability and validity. Results revealed that the measurement and structural models demonstrated an acceptable fit, and all latent constructs exhibited satisfactory reliability and construct validity. Dispositional mindfulness and self‐efficacy were found to significantly enhance psychological resilience, whereas the perceived restorativeness of nature showed no significant effect. Perceived climate resilience was positively associated with coping appraisals, life satisfaction, and positive affect, while negatively linked to eco‐emotions. The study underscores the need for further empirical investigation to validate these associations and develop a comprehensive psychological model of climate resilience. Results may inform policy development and targeted interventions aimed at strengthening psychological resilience and promoting mental health amidst escalating climate challenges, contributing to the achievement of SDG 3 (Good Health and Well‐being), SDG 11 (Sustainable Cities and Communities), and SDG 13 (Climate Action).

Conditional expectations of quantum observables: II. A causal model for the Pauli equation*

Infective endocarditis (IE) carries high in-hospital mortality, particularly among intensive care unit (ICU) patients. The predictive role of blood culture positivity in these patients remains unclear. We analyzed 484 adult IE patients from the Medical Information Mart for Intensive Care III (MIMIC-III) database, divided into training (n = 339) and testing (n = 145) cohorts. A suite of demographic, clinical, laboratory, and blood culture variables was used to develop tree-based machine learning models. Random Forest (RF) and Extreme Gradient Boosting (XGB) emerged as top performers and were combined into an ensemble model. SHapley Additive exPlanations (SHAP) quantified variable importance, while the Generative Adversarial Nets for Inference of Individualized Treatment Effects (GANITE) model assessed the average treatment effect (ATE) and conditional treatment effects (CATE) of blood culture positivity on in-hospital mortality across various clinical subgroups. The ensemble model demonstrated robust performance with an area under the receiver operating characteristic curve (AUROC) of 0.826 and an accuracy of 0.821 on the test set. Blood culture positivity consistently ranked among the top predictors of mortality. SHAP analysis revealed that the presence of bacteremia increased the predicted probability of in-hospital mortality. Specifically, the GANITE model estimated that blood culture positivity raised mortality by 0.9% (95% confidence interval [CI]: -0.9% to 2.6%) in the training set, 7.4% (95% CI: 4.3% to 10.4%) in the test set, and 2.8% (95% CI: 1.2% to 4.4%) overall. Furthermore, CATE analysis highlighted that the adverse impact of blood culture positivity was significantly more pronounced in patients aged 60 years and older, those with systolic blood pressure below 100 mmHg, and in certain endocarditis subtypes. Blood culture positivity at ICU admission is associated with a modest yet clinically significant increase in in-hospital mortality among IE patients. The application of advanced machine learning and causal inference models enhances risk stratification and may inform more targeted clinical interventions in this high-risk group.

Identifying causal relationships is essential for understanding the mechanisms through which natural and anthropogenic factors interact within Earth systems. However, in spatial cross-sectional data, the absence of temporal ordering poses significant challenges to traditional causal inference methods. This study proposes a novel Geographical Pattern Causality (GPC) model to detect positive, negative, dark causality and its strength between variables in spatial data. Grounded in dynamical systems theory and generalized embedding principles, the method transforms spatial neighbourhoods into lagged sequences, reconstructs the phase space, and compares symbolic trajectories to assess predictability and consistency in pattern changes—thereby inferring both the direction and type of causality. Case studies demonstrated that, compared to correlation analysis and Linear Non-Gaussian Acyclic Model (LiNGAM), the GPC model could reveal latent causal relationships among weakly correlated variables in geographical systems and capture diverse causal patterns. Despite limitations, such as sensitivity to noise and potential biases from proxy variables, the GPC model provides a novel framework for causal inference based on spatial observations, and it advances both the methodological and theoretical development of causality analysis in complex geographical systems.

Estimating individualized causal effects plays a vital role in data-driven decision-making, especially in high-risk domains such as public health. However, current causal inference models often lack flexibility and generalizability due to the tight coupling between representation learning and effect estimation. This study aims to develop a modular and adaptive framework to enhance the analysis of individualized causal effects in complex health data. We propose CAUSALRLSTACK, a modular framework designed to separate representation learning from causal effect estimation. In practice, the model uses a memory-augmented Transformer (TITAN) to capture complex, individualized representations. It is further paired with a doubly robust estimator(DRLearner) to improve the treatment effect estimation. A reinforcement learning agent adjusts how much each component contributes by assigning instance-specific weights. This adaptive weighting process improves the model's ability to generalize across different populations. Input features are derived from causal graphs, automatically chosen between an expert-defined graph and one discovered from data. To evaluate performance, we applied the framework to two publicly available HIV datasets that reflect community-level testing behavior and post-intervention clinical outcomes. CAUSALRLSTACK outperforms six state-of-the-art causal inference models across both datasets, achieving the highest accuracy (0.861 and 0.855), F1-Score (0.845 and 0.839), and AUC-ROC (0.897 and 0.892). It also achieves the lowest predictive uncertainty (0.093 and 0.092), indicating robust performance in estimating treatment effects. The proposed framework offers a flexible and effective solution for individualized causal inference. Its modular architecture and reinforcement learning-based weighting strategy enable adaptive, data-driven estimation across diverse populations. Strong experimental results demonstrate the potential of the framework to advance individualized causal inference in health data and provide a practical basis for designing personalized intervention strategies in HIV and broader public health domains.

Abstract Background: Analyses of etiological heterogeneity use molecular markers to sub-type cancers (e.g. breast cancer) and test whether an exposure is more strongly associated with one tumor sub-type as opposed to other sub-types. However, these molecular markers (e.g. a mutated oncogene) may best be conceptualized as mediators of exposure-disease relationships. Methods: Seven model causal scenarios, based on data from the literature on post-menopausal breast cancer, were created with varying relationships between an exposure (E), a mediating or non-mediating molecular marker (M) and a disease status (D), with M used to sub-type the disease. Using numerical examples, we assessed how well analyses of etiological heterogeneity identify the distinct etiological pathways in these seven scenarios. Results: etiological heterogeneity analyses of the data from the four causal mediation models produced similar results despite the mediating role of M differing across scenarios, with heterogeneity effect sizes ranging from 1.47 to 1.60. Analyses of data from the two scenarios in which M did not affect the risk of D, also produced evidence of etiological heterogeneity, with heterogeneity effect sizes of 1.53 and 2.66. The smallest heterogeneity effect size (1.33) was observed in the scenario where M was a sufficient cause of D. Conclusions: This methodological research indicates that analyses of etiological heterogeneity cannot distinguish between the modeled causal scenarios when a mediating or non-mediating molecular marker is used to sub-type tumors. Impact: We discuss these results in the context of the literature on obesity and etiological heterogeneity in breast tumors sub-typed by sex hormone receptor status.

Although previous experimental studies showed that metabolic acidosis promoted muscle catabolism and impaired protein synthesis, few epidemiological studies reported an independent association between serum bicarbonate levels and muscle atrophy in patients with chronic kidney disease (CKD). We examined the association between serum bicarbonate levels and low mid-upper arm circumference (MUAC), a surrogate marker of low muscle mass, in older adults with non-dialysis-dependent CKD. A total of 174 patients aged ≥65 years with an estimated glomerular filtration rate <60 ml/min/1.73 m2 (33.9% women) were eligible. We cross-sectionally examined the association between serum bicarbonate levels and MUAC using multiple linear regression, adjusting for potential confounders selected by a directed acyclic graph of our causal model. The association between serum bicarbonate and low MUAC was examined by multiple Poisson regression with robust variance. We used two cutoffs to define low serum bicarbonate: serum bicarbonate levels <24 and <22 mmol/L. Serum bicarbonate levels were positively associated with MUAC (coefficient = 0.158, 95% CI = 0.026-0.289; P = 0.019). Serum bicarbonate levels <24 mmol/L were associated with a higher prevalence of low MUAC (prevalence ratio = 3.50, 95% CI = 1.61-7.61; P = 0.002), whereas the association was attenuated for serum bicarbonate levels <22 mmol/L. Restricted cubic spline analyses suggested a nonlinear association between serum bicarbonate levels and low MUAC. We found an independent association between serum bicarbonate levels <24 mmol/L and low MUAC, calling for further prospective studies to elucidate the target serum bicarbonate level that would help retard muscle atrophy.

Background: Cardiovascular disease (CVD) is the leading cause of death among women in the US. Adverse pregnancy outcomes, particularly hypertensive disorders of pregnancy (HDP), are increasingly recognized as indicators of future CVD risk. Studies in diverse populations, especially those most affected by maternal morbidity and mortality, are urgently needed to direct guidelines and policy in the US. Objectives: To determine the association between HDP and incident CVD in a diverse, real-world population; assess mediation by postpartum cardiometabolic conditions; and identify sociodemographic and structural correlates of incident postpartum hypertension. Methods: We used linked electronic health record and survey data from the All of Us Research Program. Pregnancy episodes were identified using a validated algorithm. Cox regression and causal mediation models were used to estimate associations overall and stratified by the presence of pre-pregnancy cardiometabolic conditions (i.e., hypertension, obesity, diabetes, hyperlipidemia, and chronic kidney disease). Logistic regression was used to assess correlates of incident postpartum hypertension. Results: Participants (n = 17,673) had a mean age of 30 years [IQR: 25, 35] years; 16% identified as Black, 42% as Hispanic, 35% reported a family income &lt; $25,000, and 38% had &lt; high school education. HDP was present in 12% of pregnancies. Over a median follow-up of 4.6 years, 724 women developed incident CVD. HDP was associated with a higher incidence of CVD (12.5 vs. 7.1 per 1,000 person-years; HR 1.85, 95% CI 1.51-2.26) (Figure 1A-1B). Among women without pre-pregnancy cardiometabolic conditions, the associations were stronger. However, the absolute CVD incidence was highest among women with pre-pregnancy cardiometabolic conditions regardless of HDP status (Figure 1B-1C). Incident postpartum hypertension mediated 87% (95% CI: 45, 100) of the association between HDP and CVD. Low income, rurality, and barriers to care, such as transportation and caregiving were associated with higher odds of incident postpartum hypertension (Figure 1D). Conclusion: HDP is a significant early-life marker of premature CVD risk, especially among women without pre-pregnancy cardiometabolic conditions. Interventions that target cardiometabolic health before pregnancy and improve postpartum hypertension management, particularly among underserved populations, represent critical opportunities for CVD prevention across the life course.

The ISCHEMIA trial showed no significant difference in the rate of primary outcome events between patients with chronic coronary artery disease who used invasive strategies (INV) versus conservative strategies (CON). In this study, we used causal forest analysis to identify potentially heterogeneous treatment effects using half of the trial data.We constructed 1000 trees and the minimum number of observations of leaf nodes in each tree is 60. The remaining data were used as an independent test set using Cox proportional hazards regression modeling to test for potentially heterogeneous treatment effects. Of 5179 participants in ISCHEMIA, 1220 participants (23.5%) were LDL-C &gt; 86mg/dl and fasting plasma glucose &gt; 104mg/dl, with 595 of 1220 participants (48.8%) randomized to the training data set and 624 of 1220 participants (51.2%) randomized to the testing data set; 914 of 1220 participants (74.9%) were male, with the mean age of 63.8 ±9.5 years and 187 of 1220 (15.3%) participants experienced a primary outcome event. Using causal forest model, we identified subgroups with possible heterogeneity in INV. In the testing data, Cox models revealed that a subgroup (LDL-C＞86 mg/dl with FPG＞104 mg/dl) had a significantly higher frequency of primary outcome events in CON (18.1% [56 of 309]) than in INV (12.1% [38 of 315]), with a HR of 0.66 (95% CI, 0.44-0.99; P=0.045). Finally, we found that chronic coronary disease patients with LDL-C＞86 mg/dl and FPG＞104 mg/dl can achieve more benefit from INV, the finding that provides a new direction for individualized therapy.

BackgroundDiagnosis-related group (DRG) payment methods are increasingly being used to decrease the costs of healthcare worldwide. However, the effectiveness of cost controls varies from region to region. This study aimed to analyze the impacts of DRG payments on medical costs in China and provide theoretical support for the promotion of DRG payments in other countries.MethodsPatients from City Wuxi in China was selected, which underwent a reform from fee-for-service (FFS) payment to DRG payment during the study period. Ordinary least regression analysis (OLS) and propensity-score-matching (PSM) were used to analyze the effects of DRG, Causal Forest (CF) of machine learning algorithm was used to analyze the underlying reasons for the results.ResultsThe OLS model revealed that personal total medical costs decreased by 28.3% after the DRG reform and the total personal out-of-pocket payment (OPP) decreased by 21.3% after the DRG reform, but the personal out-of-pocket ratio increased by 15% after the DRG reform. The PSM-OLS model regression and the DRG reform results indicated decreases of 29.4% and 24.2% in personal total cost and OPP costs, respectively. The proportion of OPP costs increased by 9%. The causal forest model suggested that age and the number of surgeries played a significant role in the impact of DRG reform on patients’ medical burden (total medical expenses, OPP costs, and OPP Ratio). Results indicate that the impacts of the DRG reform was associated with a 27% reduction in patients’ medical burden (SE = 0.007), a 19.4% reduction in out-of-pocket expenses (SE = 0.012), and a 1.4% increase in utilization costs (SE = 0.002).ConclusionsDRG payment can control the growth of medical expenses and ease the burden on the medical insurance fund. However, the current rules may increase the OPP ratio and the economic burden on patients. A regulatory model in line with China’s national conditions still must be explored.

Although self-control is commonly believed to contribute to greater well-being, current evidence is inconclusive due to methodological and statistical issues. Indeed, there are both theoretical and empirical grounds to expect the opposite causal relation: wellbeing could precede self-control. We aimed to clarify this debate with two three-wave longitudinal studies, one on an Asian and the other on an American sample. We applied the random intercept cross-lagged panel model (RI-CLPM) to disentangle the stable-trait-level associations and within-persons relations between self-control and well-being. We found that earlier levels of well-being positively predicted levels of self-control 1 month (Study 2) and 6 months (Study 1) later. However, self-control did not predict later well-being. Our findings emphasize the need to reconsider the interpretations of previous—mostly between-persons—findings about self-control and well-being. Implications for understanding trait self-control, alternative causal models between self-control and well-being, and the primacy of well-being are discussed.

Can network modelling of single-cell transcriptomic data identify cellular developmental trajectories of fallopian tube (FT) epithelium and reveal functional and pathological divergence from the endometrium? A bidirectional differentiation trajectory originating from a novel OVGP1+ progenitor population of FT epithelial cells was uncovered, and causal network modelling of whole-transcriptome activity in the FT and endometrium revealed functional divergence between their secretory epithelial cells, with implications for ectopic pregnancy candidate genes. The FT forms the in vivo peri-conceptual environment, which has a significant impact on programming offspring health. The FT epithelium establishes this environment; however, the epithelial cell types are poorly characterized in health and disease. Publicly available, benign FT single-cell RNA sequencing (scRNA-seq) samples from 13 women across three previous studies were combined. Endometrial scRNA-seq samples from 13 women from one study were used to demonstrate transcriptomic differences between the epithelia of the two tissues. Network models of transcriptomic action were constructed with hypergraphs. A meta-analysis of FT scRNA-seq samples was performed to identify epithelial populations. Differential gene expression assessed differences between FT and endometrial epithelial scRNA-seq data. Functional differences between secretory cells in the tissues were characterized using hypergraph models. To identify associations with ectopic pregnancy, expression quantitative trait loci (eQTLs) from a recent GWAS were mapped onto the network models. Epithelial cells (n = 14360) were clustered into eight secretory and ciliated epithelial populations in the meta-analysis of three scRNA-seq datasets. A novel OVGP1+ epithelial progenitor cell was also identified, and its bidirectional differentiation to mature secretory or mature ciliated populations was mapped by RNA velocity analysis. This progenitor exhibited a high velocity magnitude (12.47) and low confidence (0.69): a combination strongly indicative of multipotent progenitor status. Comparing FT epithelial cells with endometrial epithelial cells revealed 5.3-fold fewer shared genes between FT and endometrial glandular secretory cells than between FT and endometrial ciliated cells, suggesting functional divergence of secretory cells along the reproductive tract. Hypergraphs were used to identify highly coordinated regions of the transcriptome robustly associated with functional gene networks. In the FT secretory cells, these networks were enriched for lipid-related (false discovery rate (FDR) < 0.002) and immune-related (FDR < 0.00007) pathways. We mapped eQTLs from a GWAS meta-analysis of 7070 women with ectopic pregnancy over a range of significance (P = 1.68 × 10-21-5.8 × 10-4) to the hypergraphs of FT and endometrium. Of the 22 genes present in the hypergraphs, 13 of these clustered as highly coordinated genes. This demonstrated the functional importance of MUC1 in the FT and endometrium (GWAS Study P = 5.32 × 10-9) and identified additional genes (SLC7A2, CLDN1, GLS, PEX6, PLXNA4, NR2F1, CLGN, PGGHG, and ANKRD36) implicated in ectopic pregnancy and eutopic pregnancy. The sample size of reproductive age women was limited in previous studies, and though causal network modelling was used and previous mechanistic data supports candidate gene involvement, no in vitro or in vivo validation of candidate was performed. These findings consolidate the existing single-cell transcriptomic datasets of the FT to provide a comprehensive understanding of epithelial populations and define functionally distinct secretory cells that contribute to the peri-conceptual environment of the FT. This study further implicates the role of MUC1 and secretory cells in ectopic pregnancy and suggests future targets for investigating embryo implantation in the FT and endometrium. No funding was received for this study. The authors do not disclose any competing interests. N/A.

Corn (Zea mays L.) yield productivity is driven by a multitude of factors, specifically genetics, environment, and management practices, along with their corresponding interactions. Despite continuous monitoring through proximal or remote sensors and advanced predictive models, understanding these complex interactions remains challenging. While predictive models are improving with regard to accurate predictions, they often fail to explain causal relationships, rendering them less interpretable than desired. Process-based or biophysical models such as the Agricultural Production Systems sIMulator (APSIM) incorporate these causalities, but the multitude of interactions are difficult to tease apart and are largely sensitive to external drivers, which often include stochastic variations. To address this limitation, we developed a novel methodology that reveals these hidden causal structures. We simulated corn production under varied conditions, including different planting dates, nitrogen fertilizer amounts, irrigation rules, soil and environmental conditions, and climate change scenarios. We then used the simulation results to rank features having the largest impact on corn yield through Random Forest modeling. The Random Forest model identified nitrogen uptake and annual transpiration as the most influential variables on corn yield, similar to the existing research. However, this analysis alone provided limited insight into how or why these features ranked highest and how the features interact with each other. Building on these results, we deployed a Causal Bayesian model, using a hybrid approach of score-based (hill climb) and constraint-based (injecting domain knowledge) models. The causal analysis provides a deeper understanding by revealing that genetics, environment, and management factors had causal impacts on nitrogen uptake and annual transpiration, which ultimately affected yield. Our methodology allows researchers and practitioners to unpack the “black box” of crop production systems, enabling more targeted and effective model development and management recommendations for optimizing corn production.

Purpose: This study investigates the influence of leadership commitment on sustainable urban development outcomes in Ghana’s construction sector, with emphasis on economic, social, and environmental dimensions of sustainability. Study Design/Methodology/Approach: A quantitative research design was adopted. A stratified random sampling method was employed to sample 430 respondents. Data were collected through structured questionnaires from Small and Medium-Sized Construction Enterprises (SMEs) across the Greater Accra, Ashanti, Central and Northern regions of Ghana. Structural Equation Modelling (SEM) was employed to examine causal relationships and model predictive strength regarding sustainability outcomes. Findings: The study found a significant positive relationship between leadership commitment and sustainable urban development outcomes in Ghana. Leadership commitment plays a crucial role in driving sustainability efforts by influencing policy formulation, resource allocation, and the effectiveness of governance. The SEM results demonstrate a strong positive relationship between leadership commitment and sustainability performance, accounting for 68% of the variance in outcomes. Research Limitation: The quantitative approach, while providing statistical rigour, may have overlooked deeper contextual insights that qualitative methods could have revealed. Social Implication: The findings reinforce the importance of leadership-driven governance in enhancing institutional accountability, policy stability, and urban resilience. Practical Implication: Strengthening leadership commitment, particularly in sustainability-oriented decision-making, can help accelerate Ghana’s transition toward sustainable cities. Originality/Value: This study provides empirical evidence on leadership commitment as a critical driver of sustainability in urban development within developing economies.

ERP-based interbrain causal model reveals closed-loop information interaction in interpersonal negotiations.

Hippocampal-based associative learning is a cornerstone of human behavior and is compromised in schizophrenia. Learning is subserved by time-driven brain network dynamics during encoding and retrieval of associations, but the impact of these dynamics on effective connectivity (EC) is unclear. We used Dynamic Causal Modeling (DCM) to investigate (a) patient-control differences in connectivity, (b) time-related impacts on connectivity, and (c) the interaction between group and time. We investigated a closed network of regions including the prefrontal and anterior cingulate cortices, the hippocampus, and regions in the magno- and parvo-cellular visual pathways. fMRI data (3 T Siemens Verio) were collected in 90 participants (52 stable schizophrenia patients) using an object-location associative learning paradigm with conditions for (a) Encoding (encoding what goes where) and (b) Retrieval (retrieving what was in the cued location). Competing network architectures were evaluated using DCM. We analyzed DCM-derived EC estimates of the contextual modulation of pathways by task condition and time, with a focus on inter-group differences and interactions with time. The winning model architecture revealed contextual modulation of bottom-up but not top-down pathways. In SCZ, encoding resulted in reduced contextual modulation of hippocampal ➔dACC and the dlPFC ➔ dACC pathways (emphasizing the importance of this triumvirate of regions). In SCZ, retrieval resulted in reduced modulation of the dlPFC ➔ dACC pathway. We observed widespread evidence for the plasticity of connection strengths over time (independent of Group). Finally, retrieval evoked an interaction in the contextual modulation of the inferior temporal ➔ hippocampus pathway (decreased modulation over time in HC, but increased modulation in SCZ). Our DCM-based investigation re-affirms the important role of frontal-hippocampal pathways in memory encoding and retrieval, elucidates the nature of memory dynamics, and reveals task-evoked disruption of network function in schizophrenia.

Farmers agronomic management responses to extreme drought and rice yields in Bihar, India

Causal Interest Modeling and Popularity Bias Mitigation in Conversational Recommender Systems

Surrounding residential greenness and mental health: Findings from the French CONSTANCES cohort.