Causal Forests Research Articles

Heterogeneous effect of sodium-glucose cotransporter 2 inhibitors on cardiovascular events among people with type 2 diabetes: an application of machine learning to target trial emulation

Abstract Background Sodium-glucose cotransporter 2 inhibitors (SGLT2i) are widely used to reduce cardiovascular risk in patients with type 2 diabetes. However, certain patient subgroups, including younger individuals or those with lower body mass index (BMI), have been underrepresented in previous randomized clinical trials. Therefore, the consistency of the effectiveness of SGLT2i has not been thoroughly investigated. Purpose This research aims to investigate the individual benefit of SGLT2i on cardiovascular risk reduction across the diverse patient population with type 2 diabetes Methods We investigated the effectiveness of SGLT2i versus active comparators (dipeptidyl peptidase 4 inhibitors [DPP4i]) using a target trial emulation framework, prevalent new-user design, and a nationwide insurer-based database covering more than 30 million working-age citizens in Japan. Participants were identified using data from April 2015 to March 2022 and were followed until March 2023. The primary endpoint was the composite of all-cause death, myocardial infarction, stroke, or heart failure. We then applied the machine learning causal forest algorithm to the emulated trial cohort to predict the individualized benefit of SGLT2i in reducing cardiovascular outcomes. After describing the characteristics of patients with predicted benefits above the median (high-benefit group) and those with predicted benefits below the median (low-benefit group), we compared the long-term survival outcomes between the high-benefit group and the low-benefit group using adjusted Cox proportional hazard models. Results The study included 274,886 participants (SGLT2i, n=137,443; DPP4i, n=137,443) with a mean age of 55.7 years, 17.3% (n=47,517) female, and a mean BMI of 27.3kg/m2. Over a median follow-up of 36 months, 3.6% (n=9,896) experienced the primary outcome. SGLT2i were associated with a reduced incidence of the primary endpoint in the entire cohort (hazard ratio [HR] = 0.94 [95% confidence interval (95%CI): 0.90–0.97]). The causal forest model showed that the effects of SGLT2i were heterogeneous across individuals. Patients in the low-benefit group were characterized by older age, female gender, lower BMI, lower blood pressure, worse kidney function, and milder diabetes status than those in the high-benefit group. While we found the preventive effect of SGLT2i on long-term cardiovascular outcomes in the high-benefit group (HR [95%CI] = 0.86 [0.82–0.90]), the effect was not observed in the low-benefit group (HR [95%CI] = 1.05 [1.00–1.11]; p-for-interaction < 0.001). Conclusion By applying a machine-learning causal forest to a target trial emulation cohort in the general population, we found heterogeneity in the effect of SGLT2i on cardiovascular outcomes, and identified characteristics of patients who are less likely to benefit from SGLT2i. The findings suggest personalized treatment strategies for cardiovascular risk reduction in patients with type 2 diabetes.Graphical summary

Application of artificial intelligence to study the causality in public health: a systematic review

Abstract Introduction In epidemiology, establishing causation requires meeting rigorous criteria beyond mere association. Traditional methodologies often struggle to provide definitive causal conclusions due to confounding variables. However, leveraging artificial Intelligence (AI) and machine Learning (ML) offers a promising solution by modeling complex interactions among relevant factors. This systematic review explores the AI techniques utilized to uncover causal relationships in public health applications. Methods A literature search was conducted in Pubmed, Web of Science, and Scopus, employing the following search strategy: (Causalit*[Title/abstract] OR Causation*[Title/abstract] OR Causal[Title/abstract]) AND (‘Machine Learning’[Title/abstract] OR “Deep Learning” [Title/abstract] OR ‘Artificial Intelligence’ [Title/abstract] OR Algorithm* [Title/abstract]). The search aimed to identify studies encompassing causal inference techniques utilizing ML and AI methodologies in public health. Results From a total of 28,230 articles published up to August 2023, after removing duplicates (11,158) and articles that did not meet the inclusion criteria (16,168 post title/abstract screening, and 856 post full-text screening), the systematic review included 48 articles. Bayesian additive regression trees, Bayesian network models, causal forests analysis, and graphical causal models emerged as the prevailing causal methodologies. AI-driven causal methodologies have been investigated across various domains within public health, encompassing disease prediction, evaluation of treatment efficacy, identification of risk factors, analysis of health behaviors, precision Public Health initiatives, and environmental health assessment. Discussion While our systematic review affirms the potential of AI-driven causal approaches across several public health domains, further research is warranted before these emerging techniques can be effectively integrated into real settings. Key messages • In epidemiology, establishing causation requires meeting rigorous criteria beyond mere association and traditional methodologies often struggle to provide definitive causal conclusions. • This systematic review explores the AI techniques utilized to uncover causal relationships in public health applications.

Impact of addressing modifiable risk factors on the 10-year risk of cardiovascular disease in individuals with familial hypercholesterolemia: a causal analysis utilising UK biobank

Abstract Background Familial hypercholesterolemia (FH) is a prevalent autosomal dominant disorder characterised by elevated low-density lipoprotein cholesterol (LDL-C) levels from birth. These elevated LDL-C levels significantly amplify the lifetime risk of cardiovascular disease (CVD). While the impact of early and effective initiation of lipid-lowering medication (LLM) on CVD risk is well documented in genetically confirmed FH, the additional impact of addressing modifiable lifestyle risk factors (MLRF) in that patient population remains largely unquantified. Methods This study quantified the potential effects of successfully managing 4 MLRF (smoking, obesity, alcohol consumption, and low physical activity) on the 10-year incidence of a first CVD diagnosis in adults with genetically confirmed FH treated with LLM within the UK Biobank. The synthesised causal estimates of these 4 MLF were obtained using two recent causal analysis frameworks, namely the doWhy causal calculus framework and the Causal Forest Double Machine Learning Method. Results In 660 FH individuals (mean age 56±8 years; 60.15% women), smoking cessation and reducing body mass index below 30 kg/m2 delivered the most substantial risk reductions (pooled estimates, adjusted for the 3 other MLRF: -6.5% and -3.2%, respectively; both p<0.001), followed by initiating moderate or high physical activity (-1.7%; p<0.001; Figure). Notably, cessation of alcohol consumption significantly increased CVD risk (+2.5%; p<0.001), aligning with existing literature. Conclusion In over 10 years, reductions in CVD risk ranging from 1.7% to 6.5% were achievable through smoking cessation, BMI reduction, and enhanced physical activity in FH individuals. Leveraging causal inference methodologies facilitates the discernment of which risk factors offer the most pronounced additional benefits beyond lipid-lowering interventions in this high-risk population.Forest Plot of Causal Effect Estimates

Do non-farmers pay more for land than farmers?

Abstract The increase in farmland prices in many parts of the world over the past decade has sparked discussions about whether non-farmers pay higher prices for farmland. This study uses a causal machine learning approach with a rich data set of land transactions in Germany to quantify the potential price premium paid by non-farmers. By applying the causal forest method, we uncover the heterogeneity of price premiums and reveal moderating effects of covariates. We find that the average positive price premium by non-farmers decreases with parcel size and distance to a highway exit.

Unveiling the causal effects of China’s minimum living standard guarantee on household transportation expenditures: A causal forest analysis

We investigate the causal effects of China’s Minimum Living Standard Guarantee (MLSG, also called Dibao) subsidy on household transportation expenditures. Utilizing data from the Chinese Social Survey, we employ a combination of the propensity score matching (PSM), causal forest (CF) and marginal treatment effect model (MTE) methodologies to rigorously estimate the subsidy’s effects. The PSM allows us to mitigate selection bias by matching MLSG recipients with comparable non-recipients. While the causal forest captures the heterogeneity of treatment effects across various household profiles. The result of MTE indicate that observable and essential heterogeneity are present to influence the effect of their subsidies, which present the consistent with PSM and CF. The causal mediation analysis indicates the mediating mechanism that MLSG impacts on household transportation expenditures, while also revealing significant variations among different regions. The study not only refines our understanding of the MLSG’s effects on household spending but also offers novel insights into applicational advancements by incorporating machine-learning techniques for policy evaluation. These results have important implications for policy formulation and refinement, particularly in the urban-rural differences.

Evaluating multimedia advertising campaign effectiveness

Companies increasingly combine multiple media outlets when launching advertising campaigns. This study employs causal forest to examine the effects of complex multimedia campaigns. The model effectively corrects for selection bias, automatically identifies informative consumer features, and performs automated data-driven consumer segmentation based on the consumer features identified. We analyze a large dataset involving around seven million consumers and four thousand covariates, and provide empirical evidence on the nonlinear effect of repeated ad exposures in the multimedia context, how such effect varies across consumer groups, and the contingent existence of multimedia synergy. We demonstrate that negligence of the selection bias and heterogeneity across segments results in suboptimal conversions and a waste of advertising resources. The analysis procedure that we propose can facilitate decision making for complex advertising campaigns to improve their effectiveness.

Heterogeneous effects of Medicaid coverage on cardiovascular risk factors: secondary analysis of randomized controlled trial

ObjectivesTo investigate whether health insurance generated improvements in cardiovascular risk factors (blood pressure and hemoglobin A1c (HbA1c) levels) for identifiable subpopulations, and using machine learning to identify characteristics of people...

Machine learning reveals heterogeneous associations between environmental factors and cardiometabolic diseases across polygenic risk scores

BackgroundAlthough polygenic risk scores (PRSs) are expected to be helpful in precision medicine, it remains unclear whether high-PRS groups are more likely to benefit from preventive interventions for diseases. Recent methodological advancements enable us to predict treatment effects at the individual level.MethodsWe employed causal forest to explore the relationship between PRSs and individual risk of diseases associated with certain environmental factors. Following simulations illustrating its performance, we applied our approach to investigate the individual risk of cardiometabolic diseases, including coronary artery diseases (CAD) and type 2 diabetes (T2D), associated with obesity and smoking among individuals from UK Biobank (UKB; n = 369,942) and BioBank Japan (BBJ; n = 149,421).ResultsHere we find the heterogeneous association of obesity and smoking with diseases across PRS values, complicated by the multi-dimensional combination of individual characteristics such as age and sex. The highest positive correlations of PRSs and the exposure-related disease risks are observed between obesity and T2D in UKB and between smoking and CAD in BBJ (Spearman’s ρ = 0.61 and 0.32, respectively). However, most relationships are weak or negative, suggesting that high-PRS groups will not necessarily benefit most from environmental factor prevention.ConclusionsOur study highlights the importance of individual-level prediction of disease risks associated with target exposure in precision medicine.

Machine learning approaches to evaluate heterogeneous treatment effects in randomized controlled trials: a scoping review

Background and ObjectivesEstimating heterogeneous treatment effects (HTEs) in randomized controlled trials (RCTs) has received substantial attention recently. This has led to the development of several statistical and machine learning (ML) algorithms to assess HTEs through identifying individualized treatment effects. However, a comprehensive review of these algorithms is lacking. We thus aimed to catalog and outline currently available statistical and ML methods for identifying HTEs via effect modeling using clinical RCT data and summarize how they have been applied in practice. Study Design and SettingWe performed a scoping review using prespecified search terms in MEDLINE and Embase, aiming to identify studies that assessed HTEs using advanced statistical and ML methods in RCT data published from 2010 to 2022. ResultsAmong a total of 32 studies identified in the review, 17 studies applied existing algorithms to RCT data, and 15 extended existing algorithms or proposed new algorithms. Applied algorithms included penalized regression, causal forest, Bayesian causal forest, and other metalearner frameworks. Of these methods, causal forest was the most frequently used (7 studies) followed by Bayesian causal forest (4 studies). Most applications were in cardiology (6 studies), followed by psychiatry (4 studies). We provide example R codes in simulated data to illustrate how to implement these algorithms. ConclusionThis review identified and outlined various algorithms currently used to identify HTEs and individualized treatment effects in RCT data. Given the increasing availability of new algorithms, analysts should carefully select them after examining model performance and considering how the models will be used in practice.

High-dimensional Iterative Causal Forest (hdiCF) for Subgroup Identification Using Health Care Claims Data.

We recently developed a machine-learning subgrouping algorithm, iterative causal forest (iCF), to identify subgroups with heterogeneous treatment effects (HTEs) using predefined covariates. However, such predefined covariates may miss or poorly define important features leading to inaccurate subgrouping. To address such limitations, we developed a new semi-automatic subgrouping algorithm, hdiCF, which adapts methodology from high-dimensional propensity score for feature recognition in claims data. The hdiCF algorithm has 3 steps: 1) high-dimensional feature identification by International Classification of Diseases, Current Procedural Terminology, and Anatomical Therapeutic Chemical codes (in/outpatient diagnoses, procedures, prescriptions) and creation of ordinal variables by frequency of occurrence; 2) propensity score trimming and high-dimensional feature preparation; 3) iCF implementation to identify subgroups. We applied hdiCF in a 20% random sample of fee-for-service Medicare beneficiaries who initiated sodium-glucose cotransporter-2 inhibitors (SGLT2i) or glucagon-like peptide-1 receptor agonists to identify subgroups with HTEs for incidence of hospitalized heart failure. HdiCF findings were consistent with studies suggesting SGLT2i to be more beneficial for patients with pre-existing heart failure or chronic kidney disease. HdiCF is not dependent on prior hypotheses about HTEs and identifies subgroups with markers for potential HTEs in real-world evidence studies where active-comparator, new-user study designs limit the potential for unmeasured confounding.

Does Increasing Concentration Hit Poorer Areas More? A Study of Retail Petroleum Markets*

A central tenet in the field of industrial organisation is that increasing/decreasing market concentration is associated with increased/reduced markups. But does this variation affect every consumer to the same extent? Previous literature finds price dispersion exists even for homogeneous goods, at least partially as a result of heterogeneity in consumer engagement with the market. We study this question by linking demographic and income heterogeneity across local areas to the impact of changing market concentration on markups. With 15 years of station‐level motor fuel price data from Western Australia and information on instances of local market exit and entry, we apply a non‐parametric causal forest approach to explore the heterogeneity in the effect of exit/entry. The paper provides evidence of the distributional effect of changing market concentration. Areas with lower income experience a larger increase in petrol stations' price margin as a result of market exit. On the other hand, entry does not benefit the same low‐income areas with a larger reduction in the margin than in high‐income areas. Policy implications include a need to further focus on increasing engagement by low‐income consumers.

Causal effect of video gaming on mental well-being in Japan 2020–2022

The widespread use of video games has raised concerns about their potential negative impact on mental well-being. Nevertheless, the empirical evidence supporting this notion is largely based on correlational studies, warranting further investigation into the causal relationship. Here we identify the causal effect of video gaming on mental well-being in Japan (2020–2022) using game console lotteries as a natural experiment. Employing approaches designed for causal inference on survey data (n = 97,602), we found that game console ownership, along with increased game play, improved mental well-being. The console ownership reduced psychological distress and improved life satisfaction by 0.1–0.6 standard deviations. Furthermore, a causal forest machine learning algorithm revealed divergent impacts between different types of console, with one showing smaller benefits for adolescents and females while the other showed larger benefits for adolescents. These findings highlight the complex impact of digital media on mental well-being and the importance of considering differential screen time effects.

Development and Validation of a Policy Tree Approach for Optimizing Intravenous Fluids in Critically Ill Patients with Sepsis and Acute Kidney Injury.

Intravenous fluids are mainstay of management of acute kidney injury (AKI) after sepsis but can cause fluid overload. Recent literature shows that restrictive fluid strategy may be beneficial in some patients with AKI, however, identifying these patients is challenging. We aimed to develop and validate a machine learning algorithm to identify patients who would benefit from a restrictive fluid strategy. We included patients with sepsis who developed AKI within 48 hours of ICU admission and defined restrictive fluid strategy as receiving <500mL fluids within 24 hours after AKI. Our primary outcome was early AKI reversal within 48 hours of AKI onset, and secondary outcomes included sustained AKI reversal and major adverse kidney events (MAKE) at discharge. We used a causal forest, a machine learning algorithm to estimate individual treatment effects and policy tree algorithm to identify patients who would benefit by restrictive fluid strategy. We developed the algorithm in MIMIC-IV and validated it in eICU database. Among 2,091 patients in the external validation cohort, policy tree recommended restrictive fluids for 88.2%. Among these, patients who received restrictive fluids demonstrated significantly higher rate of early AKI reversal (48.2% vs 39.6%, p<0.001), sustained AKI reversal (36.7% vs 27.4%, p<0.001) and lower rates of MAKE by discharge (29.3% vs 35.1%, p=0.019). These results were consistent in adjusted analysis. Policy tree based on causal machine learning can identify septic patients with AKI who benefit from a restrictive fluid strategy. This approach needs to be validated in prospective trials.

Glucagon-like Peptide 1 Receptor Agonists and Asthma Exacerbations: Which Patients Benefit Most?

While recent evidence suggested glucagon-like peptide 1 receptor agonists (GLP1RA) may reduce the risk of asthma exacerbations, it remains unclear which subpopulations might derive the most benefit from GLP1RA treatment. To identify characteristics of patients with asthma that predict who might benefit the most from GLP1RA treatment using real-world data. We implemented an active-comparator, new-user design analysis using commercially insured patients ages 18-65 from Marketscan data 2007-2019 and identified two cohorts: GLP1RA vs thiazolidinediones and GLP1RA vs sulfonylureas. The outcome was acute exacerbation of asthma (hospital admission or emergency department (ED) visit for asthma) within 180 days after initiation. We applied iterative causal forest (iCF), a novel causal machine learning subgrouping algorithm, to assess HTE. In identified subgroups, we predicted propensity score, conducted propensity score trimming, and then estimated adjusted risk differences (aRD) for the effect of GLP1RA relative to comparators on asthma exacerbation using inverse probability treatment weighting in propensity score trimmed subpopulation. Among 10,989 patients initiating GLP1RA or thiazolidinediones and 17,088 patients initiating GLP1RA vs sulfonylurea, GLP1RA initiators had fewer exacerbations with an aRD of -0.5% (95% CI -1.1% to 0.1%) and -1.6% (95% CI -2.2% to -1.1%), respectively. In the GLP1RAvsSUcohort where we observed beneficial effect, our iCF analysis identified 5 subgroups with different treatment effects, defined by number of ED visits, number of prescriptions of short-acting beta2-agonsit (SABA), number of prescriptions of inhaled steroid (ICS) and long-acting beta-agonists (LABA) (either combination therapy or concurrent use), and aged 50+. Among these, patients with 2+ ED visits during 12-month baseline period had the largest absolute exacerbation risk reduction, with a decrease of 2.8% for GLP1RA (95% CI: -4.8% to -0.9%). GLP1RA demonstrated beneficial effect on reducing asthma exacerbation relative to sulfonylureas. Asthma patients with 2+ ED visits (a proxy for disease severity) benefit most from GLP1RA. ED visit frequency, and number of maintenance and reliever inhalers, and age may help individualize prediction of the short-term benefit from GLP1RA on asthma exacerbation.

Renal function as an effect modifier of intensive glucose control in delaying cognitive function decline among individuals with type 2 diabetes: A revisit to the ACCORD MIND trial.

Dysglycaemia accelerates cognitive decline. Intensive glucose control may help delay or prevent cognitive function decline (CFD). We aimed to determine how patient characteristics influence the effect of intensive glucose control [glycated haemoglobin (HbA1c) <6.0%] on delaying CFD in people with type 2 diabetes. In this post-hoc analysis of 2977 type 2 diabetes participants from the ACCORD MIND trial, we applied the causal forest and causal tree algorithms to identify the effect modifier of intensive glucose control in delaying CFD from 68 variables (demographics, disease history, medications, vitals and baseline biomarkers). The exposure was intensive versus standard glucose control (HbA1c <6.0% vs. 7.0%-7.9%). The main outcome was cognitive function changes from baseline to the 40th month follow-up, which were evaluated using the digit symbol substitution test, Rey auditory verbal learning test, mini-mental state examination and Stroop test. We used Cohen's d, a measure of standardized difference, to quantify the effect size of intensive glucose control on delaying CFD. Among all the baseline characteristics, renal function was the most significant effect modifier. Participants with urinary albumin levels <0.4 mg/dl [absolute function change (AFC): 0.51 in mini-mental state examination, 95% confidence interval (CI): 0.04, 0.98, Cohen's d: 0.25] had slower CFD with intensive glucose control. Patients with preserved renal function (estimated glomerular filtration rate between 60 and 90 ml/min/1.73 m2) were associated with small benefits (AFC: 1.28 in Stroop, 95% CI: 0.28, 2.27, Cohen's d: 0.12) when undergoing intensive glucose control. Conversely, participants with an estimated glomerular filtration rate <60 ml/min/1.73 m2 (AFC: -0.57 in the Rey auditory verbal learning test, 95% CI: -1.09, -0.05, Cohen's d: -0.30) exhibited faster CFD when undergoing intensive glucose control. Participants who were <60 years old showed a significant benefit from intensive glucose control in delaying CFD (AFC: 1.08 in the digit symbol substitution test, 95% CI: 0.06, 2.10, Cohen's d: 0.13). All p < .05. Our findings linked renal function with the benefits of intensive glucose control in delaying CFD, informing personalized HbA1c goals for those with diabetes and at risk of CFD.

P-660 Finding the optimal starting dose of gonadotropins for an ovarian stimulation treatment through Machine Learning

Abstract Study question Can machine learning algorithms provide an individualized starting dose of gonadotropins to maximize the number of oocytes retrieved in patients undergoing ovarian stimulation ? Summary answer Methods of Policy Learning based on Causal Inference and Machine Learning optimize starting doses of gonadotropins with a substantial gain of oocytes under appropriate treatment. What is known already Despite attempts toward standardization, there has been no consensus about establishing the optimal starting dose of gonadotropins to be used in ovarian stimulation. In the real world, medical practice differs considerably across countries and even among fertility clinics nationally. The machine learning community has proposed a variety of models for doing this, but with limited comparisons, causal inference can provide the statistical tools to enable the best model to be chosen and allow its explicability. Study design, size, duration In a retrospective single-center study, we reviewed the data of 11,436 cycles of ovarian stimulation for IVF between 2012 and 2023. In order to estimate the causal effect of the starting dose of gonadotropins on the number of oocytes retrieved, we selected a relevant subset of confounding covariates – including age, Antral Follicle Count (AFC), Anti-Müllerian Hormone (AMH) and oestradiol (E2) rates. Participants/materials, setting, methods We used five different Machine Learning models : Linear Regression, Regression Forest, Super Learner (SL), Multi-Arm Causal Forest (MACF) and a model based on the Targeted Maximum Likelihood Estimation (TMLE). We employed these models for Policy Learning through an Outcome Regression Modeling approach. We compared the gain of oocytes obtained through the optimal policy of our different models and analyzed the variables which played a crucial role in determining the optimal dose. Main results and the role of chance We use two approaches to compare our models, the first one gives the expected average gain of oocytes under the optimal policy : in this case, the Linear Regression is the best with an average gain of 0.90 oocytes, followed by the MACF (0.80), the SL (0.58), the TMLE (0.44) and the Regression Forest (0.18). Considering the lack of heterogeneity of our database we use a second approach, an Augmented Inverse Propensity Score based value, to compare with more robustness the performance of the optimal policy of each model. In this case the MACF (an explainable model specifically designed for causal inference) is the best with a gain of 0.50 oocytes, followed by the SL (0.42), the Linear Regression (0.37), the TMLE (0.23) and the Regression Forest (0.18). In general, our models recommend lower doses, for instance 20% of the patients are recommended by the MACF to reduce their initial doses of gonadotropins by on average 100 IU. The most significant covariates in the models are the AFC, the BMI, and the rate of AMH, with p-values &amp;lt; 0.001 in the Linear Regression that has a R-squared of 0.418. Limitations, reasons for caution Full IVF history was not included in patients who had previous IVF attempts, but only data from the last previous attempt was considered. Also, the study is monocentric and more heterogeneity will be present considering other centers. Wider implications of the findings This study needs to be completed by data from different centers in France and around the world to bring diversity to the patients’ data and to the treatment prescriptions in order to improve external validity. Algorithm recommendations and their adherence need to be tested prospectively. Trial registration number not applicable

Little and often: Causal inference machine learning demonstrates the benefits of homework for improving achievement in mathematics and science

BackgroundDespite its important role in education, significant gaps remain in the literature on homework. Notably, there is a dearth of understanding regarding how homework effects vary across different subjects, how student backgrounds may moderate its effectiveness, what the optimal amount and distribution of homework is, and how the causal impact of homework can be disentangled from other associations. AimsThis study examines the different effects of homework frequency and duration on student achievement in both mathematics and science while adopting a causal inference probabilistic framework. SampleOur data consists of a nationally representative sample of 4118 Irish eighth grade students, collected as part of TIMSS 2019. MethodsWe employ an extension of a causal inference machine learning model called Bayesian Causal Forests that allows us to consider the effect of homework on achievement in mathematics and science simultaneously. By investigating the impacts of both homework frequency and duration, we discern the optimal frequency and duration for homework in both subjects. Additionally, we explore the potential moderating role of student socioeconomic backgrounds. ResultsDaily homework benefitted mathematics achievement the most, while three to four days per week was most effective for science. Short-duration assignments proved equally as effective as longer ones in both subjects. Notably, students from advantaged socioeconomic backgrounds did not gain more from homework. ConclusionsThese findings can guide policies aimed at enhancing student outcomes while promoting a balance between academic responsibilities and extracurricular activities.

Exploring Heterogeneity in Cost-Effectiveness Using Machine Learning Methods: A Case Study Using the FIRST-ABC Trial.

The aim of this study was to explore heterogeneity in the cost-effectiveness of high-flow nasal cannula (HFNC) therapy compared with continuous positive airway pressure (CPAP) in children following extubation. Using data from the FIRST-line support for Assistance in Breathing in Children (FIRST-ABC) trial, we explore heterogeneity at the individual and subgroup levels using a causal forest approach, alongside a seemingly unrelated regression (SUR) approach for comparison. FIRST-ABC is a noninferiority randomized controlled trial (ISRCTN60048867) including children in UK paediatric intensive care units, which compared HFNC with CPAP as the first-line mode of noninvasive respiratory support. In the step-down FIRST-ABC, 600 children clinically assessed to require noninvasive respiratory support were randomly assigned to HFNC and CPAP groups with 1:1 treatment allocation ratio. In this analysis, 118 patients were excluded because they did not consent to accessing their medical records, did not consent to follow-up questionnaire or did not receive respiratory support. The primary outcome of this study is the incremental net monetary benefit (INB) of HFNC compared with CPAP using a willingness-to-pay threshold of £20,000 per QALY gain. INB is calculated based on total costs and quality adjusted life years (QALYs) at 6 months. The findings suggest modest heterogeneity in cost-effectiveness of HFNC compared with CPAP at the subgroup level, while greater heterogeneity is detected at the individual level. The estimated overall INB of HFNC is smaller than the INB for patients with better baseline status suggesting that HFNC can be more cost-effective among less severely ill patients.

Estimating the Heterogeneous Causal Effects of Parent-Child Relationships among Chinese Children with Oppositional Defiant Symptoms: A Machine Learning Approach.

Oppositional defiant symptoms are some of the most common developmental symptoms in children and adolescents with and without oppositional defiant disorder. Research has addressed the close association of the parent-child relationship (PCR) with oppositional defiant symptoms. However, it is necessary to further investigate the underlying mechanism for forming targeted intervention strategies. By using a machine learning-based causal forest (CF) model, we investigated the heterogeneous causal effects of the PCR on oppositional defiant symptoms in children in Chinese elementary schools. Based on the PCR improvement in two consecutive years, 423 children were divided into improved and control groups. The assessment of oppositional defiant symptoms (AODS) in the second year was set as the dependent variable. Additionally, several factors based on the multilevel family model and the baseline AODS in the first year were included as covariates. Consistent with expectations, the CF model showed a significant causal effect between the PCR and oppositional defiant symptoms in the samples. Moreover, the causality exhibited heterogeneity. The causal effect was greater in those children with higher baseline AODS, a worse family atmosphere, and lower emotion regulation abilities in themselves or their parents. Conversely, the parenting style played a positive role in causality. These findings enhance our understanding of how the PCR contributes to the development of oppositional defiant symptoms conditioned by factors from a multilevel family system. The heterogeneous causality in the observation data, established using the machine learning approach, could be helpful in forming personalized family-oriented intervention strategies for children with oppositional defiant symptoms.

Causal Forest Machine Learning Analysis of Parkinson's Disease in Resting-State Functional Magnetic Resonance Imaging.

In recent years, Artificial Intelligence has been used to assist healthcare professionals in detecting and diagnosing neurodegenerative diseases. In this study, we propose a methodology to analyze functional Magnetic Resonance Imaging signals and perform classification between Parkinson's disease patients and healthy participants using Machine Learning algorithms. In addition, the proposed approach provides insights into the brain regions affected by the disease. The functional Magnetic Resonance Imaging from the PPMI and 1000-FCP datasets were pre-processed to extract time series from 200 brain regions per participant, resulting in 11,600 features. Causal Forest and Wrapper Feature Subset Selection algorithms were used for dimensionality reduction, resulting in a subset of features based on their heterogeneity and association with the disease. We utilized Logistic Regression and XGBoost algorithms to perform PD detection, achieving 97.6% accuracy, 97.5% F1 score, 97.9% precision, and 97.7%recall by analyzing sets with fewer than 300 features in a population including men and women. Finally, Multiple Correspondence Analysis was employed to visualize the relationships between brain regions and each group (women with Parkinson, female controls, men with Parkinson, male controls). Associations between the Unified Parkinson's Disease Rating Scale questionnaire results and affected brain regions in different groups were also obtained to show another use case of the methodology. This work proposes a methodology to (1) classify patients and controls with Machine Learning and Causal Forest algorithm and (2) visualize associations between brain regions and groups, providing high-accuracy classification and enhanced interpretability of the correlation between specific brain regions and the disease across different groups.