Specification Curve Analysis Research Articles

Machine learning identification of risk factors for ischaemic stroke in patients with heart failure with reduced ejection fraction but without atrial fibrillation: A report from WARCEF trial

Abstract Background The prediction of ischaemic stroke in patients with heart failure with reduced ejection fraction (HFrEF) but without atrial fibrillation (AF) remains challenging and risk stratification based clinical factors alone has limited discrimination. Our aim was to evaluate the performance of machine learning (ML) to identify risk factors for ischaemic stroke in such HFrEF patients. Methods We performed a post-hoc analysis of the WARCEF trial, only including patients without prior history of AF, and censoring those who developed incident AF during follow-up. We evaluated the performance of 9 ML models to identify risk factors for incident stroke, using metrics including area under the curve (AUC), precision, specificity, and decision curve analysis. The importance of each feature in constructing the model was quantified using SHapley Additive exPlanations (SHAP) values. Results We included 2,213 patients with HFrEF but without AF (mean age 59±12 years; 79% male). Of these, 74 (3.3%) had an ischaemic stroke during mean follow-up of 3.5±1.8 years. Out of the 29 patient-demographics variables, 12 were selected for the ML training. All ML models demonstrated high AUC values, outperforming the CHA2DS2-VASc score (Figure 1). The logistic regression model exhibited a slight deviation from this trend. The XGBoost model achieved an AUC of 0.86 (95% confidence interval [CI]: 0.72-0.99), while the Random Forest model showed an AUC of 0.85 (95% CI: 0.74-0.96). Similarly, the Support Vector Machine and Light Gradient Boosting Machine models recorded AUCs of 0.85 (95% CI: 0.72-0.97) and 0.83 (95% CI: 0.70-0.96), respectively. Each of these models attained precision and specificity scores of 1.0, indicating high performance. These models consistently provided significant net clinical benefits. Key risk factors for stroke identified across these models were creatinine clearance and blood urea nitrogen. The use of warfarin appears to be protective. Conclusions Machine-learning models can identify risk factors for ischaemic stroke in patients with HFrEF but without AF. Worsening renal failure was predictive of an incident stroke, while warfarin use was protective against stroke (Figure 2).Figure 1Figure 2

The Validation of the 2023 ACR/EULAR Antiphospholipid Syndrome Classification Criteria in a Cohort from Turkey.

Our aim was to validate the performance of the American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) classification criteria for antiphospholipid syndrome (APS), published in 2023, in an APS cohort. A total of 193 patients, 83 with APS (secondary APS, n = 45; primary APS, n = 38) and 110 without APS (systemic lupus erythematosus (SLE), n = 100; others, n = 10), were included in this study. The performance (sensitivity, specificity and area under the curve (AUC)) of the 2023 ACR/EULAR classification criteria for APS was evaluated and the agreement with the revised Sapporo criteria was compared using the kappa test. In our cohort, the sensitivity and specificity of the 2023 ACR/EULAR classification criteria for APS were 73% and 94%, respectively (AUC: 0.836, 95% CI: 0.772-0.899), while the sensitivity and specificity of the revised Sapporo criteria were 66% and 98%, respectively (95% CI: 0.756-0.888). The performance of the two sets of criteria in our cohort was significantly consistent and significant (p < 0.001). When the sensitivity, specificity and ROC curve analysis were performed again by excluding livedo racemosa, the sensitivity of the new criteria in our cohort was 62% and the specificity was 100% (AUC: 0.813, 95% CI: 0.746-0.881). Although the newly published criteria broaden the scope of APS classification by including clinical findings other than thrombosis and obstetric criteria, their sensitivity in our cohort was low. On the other hand, we found that the specificity of the criteria in our cohort reached 100% when livedo findings were excluded.

Social media and youth mental health: Simple narratives produce biased interpretations.

Many academics and pundits contend that social media use is the primary cause of an international youth mental health crisis. However, these claims often rely on correlational evidence, ignoring the confounding effects of developmental, environmental, social, and psychological factors that influence mental health. This oversimplifies the complex etiology of mental health problems. We call for a more nuanced understanding of the role of social media in youth mental health that avoids oversimplification. Additionally, we urge researchers to move beyond vague, narrative-driven verbal theories and encode them into precise, testable causal models. Using simulation techniques and specification curve analyses, we show how misspecified models that ignore these confounding factors can lead to biased conclusions about social media's adverse effects. This simplistic focus on social media use diverts attention from the broader factors contributing to youth mental health problems, hindering the development of effective interventions and support. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Individual differences in changes in subjective well-being: The role of event characteristics after negative life events.

Negative life events can lead to lasting changes in subjective well-being (SWB). However, people change differently in their SWB after negative life events, and our understanding of factors explaining these individual differences is still limited-possibly because research so far has neglected to investigate differences in the characteristics of the experienced events (e.g., perceived impact, causes of the event). To address this gap, we examined whether perceived event characteristics and objective-descriptive characteristics of negative life events can explain individual differences in changes in SWB. We used data from a longitudinal study in which the SWB of participants (N = 1,068) who had recently experienced a negative life event was assessed at five measurement occasions over 6 months. Perceived event characteristics and objective-descriptive event characteristics were significantly related to each other. Furthermore, both kinds of event characteristics were associated with individual differences in changes in SWB. Finally, specification curve analyses illustrated that several analytical decisions (e.g., the examined SWB component) influenced the association between an event characteristic and changes in SWB. Results from these specification curve analyses can be accessed via a ShinyApp (https://life-event-research.shinyapps.io/EventCharacteristics/). Our findings provide insights into possible causes of the event perception and show that both perceived event characteristics and objective-descriptive event characteristics can help to better understand individual differences in the reaction to major life events. However, as effects seem to depend on several analytical decisions, future research is needed to identify the important characteristics of life events for different events and outcomes. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

The etiology of the association between parental nurturance and youth antisocial behavior: Evidence from a twin differences study

AbstractBackgroundLower parental nurturance is consistently associated with higher levels of youth antisocial behavior (ASB), but the etiology of this association remains unclear. To fill this gap, we employed a twin differences approach to illuminate the environmental and genetic origins of the association between parental nurturance and children's ASB.MethodsParticipants were 2060 twins (49% female) ages 6–10 from the Michigan State University Twin Registry. Parental nurturance and youth ASB were assessed using multiple measures (e.g., questionnaires, interviews) and informant reports (e.g., twins, parents, teachers). Co‐twin difference‐score correlations were analyzed separately by zygosity using specification curve analysis, an exhaustive modeling approach that examined associations across all possible specifications of the nurturance and ASB data.ResultsParental nurturance demonstrated clear, negatively signed associations with youth ASB at the individual level. However, these associations generally did not persist within twin pairs. We observed no significant twin difference correlations within monozygotic (MZ) pairs and only a handful of significant twin difference correlations among dizygotic (DZ) pairs, in which the DZ co‐twin who experienced more nurturance exhibited less ASB. Post‐hoc analyses in these data revealed that these associations differed markedly from those with harsh parenting that suggested environmental influences on youth ASB.ConclusionsThese results strongly argue against a causal influence of low parental nurturance on youth ASB, and instead suggest that genetic influences and shared environmental confounds underlie their association. Further, findings strongly suggest that different parenting behaviors are associated with child ASB via different etiologic mechanisms.

Gravity Models in International Trade: A Specification Curve Analysis

Gravity Models in International Trade: A Specification Curve Analysis

Prediction of metastases in confusing mediastinal lymph nodes based on flourine-18 fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) imaging using machine learning.

For patient management and prognosis, accurate assessment of mediastinal lymph node (LN) status is essential. This study aimed to use machine learning approaches to assess the status of confusing LNs in the mediastinum using positron emission tomography/computed tomography (PET/CT) images; the results were then compared with the diagnostic conclusions of nuclear medicine physicians. A total of 509 confusing mediastinal LNs that had undergone pathological assessment or follow-up from 320 patients from three centres were retrospectively included in the study. LNs from centres I and II were randomised into a training cohort (N=324) and an internal validation cohort (N=81), while those from centre III patients formed an external validation cohort (N=104). Various parameters measured from PET and CT images and extracted radiomics and deep learning features were used to construct PET/CT-parameter, radiomics, and deep learning models, respectively. Model performance was compared with the diagnostic results of nuclear medicine physicians using the area under the curve (AUC), sensitivity, specificity, and decision curve analysis (DCA). The coupled model of gradient boosting decision tree-logistic regression (GBDT-LR) incorporating radiomic features showed AUCs of 92.2% [95% confidence interval (CI), 0.890-0.953], 84.6% (95% CI, 0.761-0.930) and 84.6% (95% CI, 0.770-0.922) across the three cohorts. It significantly outperformed the deep learning model, the parametric PET/CT model and the physician's diagnosis. DCA demonstrated the clinical usefulness of the GBDT-LR model. The presented GBDT-LR model performed well in evaluating confusing mediastinal LNs in both internal and external validation sets. It not only crossed radiometric features but also avoided overfitting.

The efficacy of CBCT-based radiomics techniques in differentiating between conventional and unicystic ameloblastoma

ObjectiveThe aim of this study was to develop a cone beam computed tomography (CBCT) radiomics-based model that differentiates between conventional and unicystic ameloblastoma (AB). MethodsIn this retrospective study, CBCT images were collected from 100 patients who had ABs that were diagnosed histopathologically as conventional or unicystic AB after surgical treatment. The patients were randomly divided into training (70) and validation (30) cohorts. Radiomics features were extracted from the images, and the optimal features were incorporated into five models: Logistic Regression, Support Vector Machine, Linear Discriminant Analysis, Random Forest, and XGBoost for prediction of tumor type. Model performance was evaluated using the area under the curve (AUC) from receiver operating characteristic analysis, sensitivity, specificity, accuracy, calibration curves, and decision curve analysis (DCA). ResultsThe 20 optimal radiomics features were incorporated into the Logistic Regression (LR) model, which exhibited the best overall performance with AUC = 0.936 (95% CI, 0.877-0.996) for the training cohort and AUC = 0.929 (95%CI, 0.832-1.000) for the validation cohort. The nomogram combined the clinical features and the radiomics signature and resulted in the best predictive performance. ConclusionsThe LR model demonstrated the ability of radiomics and the nomogram to distinguish between the two types of AB and may have the potential to replace biopsies under noninvasive conditions.

Development of a risk prediction score for screening for HBV, HCV and HIV among migrants in France: results from a multicentre observational study (STRADA study)

ObjectivesMigrants from high HIV, hepatitis B virus (HBV) or hepatitis C virus (HCV) endemicity regions have a great burden of these infections and related diseases in the host countries. This...

Specification curve analysis to identify heterogeneity in risk factors for dementia: findings from the UK Biobank

BackgroundIn 2020, the Lancet Commission identified 12 risk factors as priorities for prevention of dementia, and other studies identified APOE e4/e4 genotype and family history of Alzheimer’s disease strongly associated with dementia outcomes; however, it is unclear how robust these relationships are across dementia subtypes and analytic scenarios. Specification curve analysis (SCA) is a new tool to probe how plausible analytical scenarios influence outcomes.MethodsWe evaluated the heterogeneity of odds ratios for 12 risk factors reported from the Lancet 2020 report and two additional strong associated non-modifiable factors (APOE e4/e4 genotype and family history of Alzheimer’s disease) with dementia outcomes across 450,707 UK Biobank participants using SCA with 5357 specifications across dementia subtypes (outcomes) and analytic models (e.g., standard demographic covariates such as age or sex and/or 14 correlated risk factors).ResultsSCA revealed variable dementia risks by subtype and age, with associations for TBI and APOE e4/e4 robust to model specification; in contrast, diabetes showed fluctuating links with dementia subtypes. We found that unattributed dementia participants had similar risk factor profiles to participants with defined subtypes.ConclusionsWe observed heterogeneity in the risk of dementia, and estimates of risk were influenced by the inclusion of a combination of other modifiable risk factors; non-modifiable demographic factors had a minimal role in analytic heterogeneity. Future studies should report multiple plausible analytic scenarios to test the robustness of their association. Considering these combinations of risk factors could be advantageous for the clinical development and evaluation of novel screening models for different types of dementia.

Incidence rate of occult lymph node metastasis in clinical T1−2N0M0 small cell lung cancer patients and radiomic prediction based on contrast-enhanced CT imaging: a multicenter study

BackgroundThis study aimed to explore the incidence of occult lymph node metastasis (OLM) in clinical T1 − 2N0M0 (cT1 − 2N0M0) small cell lung cancer (SCLC) patients and develop machine learning prediction models using preoperative intratumoral and peritumoral contrast-enhanced CT-based radiomic data.MethodsBy conducting a retrospective analysis involving 242 eligible patients from 4 centeres, we determined the incidence of OLM in cT1 − 2N0M0 SCLC patients. For each lesion, two ROIs were defined using the gross tumour volume (GTV) and peritumoral volume 15 mm around the tumour (PTV). By extracting a comprehensive set of 1595 enhanced CT-based radiomic features individually from the GTV and PTV, five models were constucted and we rigorously evaluated the model performance using various metrics, including the area under the curve (AUC), accuracy, sensitivity, specificity, calibration curve, and decision curve analysis (DCA). For enhanced clinical applicability, we formulated a nomogram that integrates clinical parameters and the rad_score (GTV and PTV).ResultsThe initial investigation revealed a 33.9% OLM positivity rate in cT1 − 2N0M0 SCLC patients. Our combined model, which incorporates three radiomic features from the GTV and PTV, along with two clinical parameters (smoking status and shape), exhibited robust predictive capabilities. With a peak AUC value of 0.772 in the external validation cohort, the model outperformed the alternative models. The nomogram significantly enhanced diagnostic precision for radiologists and added substantial value to the clinical decision-making process for cT1 − 2N0M0 SCLC patients.ConclusionsThe incidence of OLM in SCLC patients surpassed that in non-small cell lung cancer patients. The combined model demonstrated a notable generalization effect, effectively distinguishing between positive and negative OLMs in a noninvasive manner, thereby guiding individualized clinical decisions for patients with cT1 − 2N0M0 SCLC.

Convergence of Age Differences in Risk Preference, Impulsivity, and Self-Control: A Multiverse Analysis.

Numerous theories exist regarding age differences in risk preference and related constructs, yet many of them offer conflicting predictions and fail to consider convergence between measurement modalities or constructs. To pave the way for conceptual clarification and theoretical refinement, in this preregistered study we aimed to comprehensively examine age effects on risk preference, impulsivity, and self-control using different measurement modalities, and to assess their convergence. We collected a large battery of self-report, informant report, behavioral, hormone, and neuroimaging measures from a cross-sectional sample of 148 (55% female) healthy human participants between 16 and 81 years (mean age = 46 years, standard deviation [SD] = 19). We used an extended sample of 182 participants (54% female, mean age = 46 years, SD = 19) for robustness checks concerning the results from self-reports, informant reports, and behavioral measures. For our main analysis, we performed specification curve analyses to visualize and estimate the convergence between the different modalities and constructs. Our multiverse analysis approach revealed convergent results for risk preference, impulsivity, and self-control from self- and informant reports, suggesting a negative effect of age. For behavioral, hormonal, and neuroimaging outcomes, age effects were mostly absent. Our findings call for conceptual clarification and improved operationalization to capture the putative mechanisms underlying age-related differences in risk preference and related constructs.

Beyond Prostate Imaging Reporting and Data System: Combining Magnetic Resonance Imaging Prostate Imaging Reporting and Data System and Prostate-Specific Membrane Antigen-Positron Emission Tomography/Computed Tomography PRIMARY Score in a Composite (P) Score for More Accurate Diagnosis of Clinically Significant Prostate Cancer.

The Prostate Imaging Reporting and Data System (PI-RADS) score is standard of care for clinically significant prostate cancer (csPCa) diagnosis. The PRIMARY score (prostate-specific membrane antigen [PSMA]-positron emission tomography [PET]/CT) also has high diagnostic accuracy for csPCa. This study aimed to develop an easily calculated combined (P) score for csPCa detection (International Society of Urological Pathology [ISUP] ≥2) incorporating separately read PI-RADS and PRIMARY scores, with external validation. Two datasets of men with suspected PCa, no prior biopsy, recent MRI and 68Ga-PSMA-11-PET/CT, and subsequent transperineal biopsy were evaluated. These included the development sample (n = 291, 56% csPCa) a prospective trial and the validation sample (n = 227, 67% csPCa) a multicenter retrospective database. Primary outcome was detection of csPCa (ISUP ≥2), with ISUP ≥ 3 cancer detection a secondary outcome. Score performance was evaluated by area under the curve, sensitivity, specificity, and decision curve analysis. The 5-point combined (P) score was developed in a prospective dataset. In the validation dataset, csPCa was identified in 0%, 20%, 52%, 96%, and 100% for P score 1 to 5. The area under the curve was 0.93 (95% CI: 0.90-0.96), higher than PI-RADS 0.89 (95% CI: 0.85-0.93, P = .039) and PRIMARY score alone 0.84 (95% CI: 0.79-0.89, P < .001). Splitting scores at 1/2 (negative) vs 3/4/5 (positive), P score sensitivity was 94% (95% CI: 89-97) compared to PI-RADS 89% (95% CI: 83-93) and PRIMARY score 86% (95% CI: 79-91). For ISUP ≥ 3, P score sensitivity was 99% (95% CI: 95-100) vs 94% (95% CI: 88-98) and 92% (95% CI: 85-97) for PI-RADS and PRIMARY scores respectively. A maximum standardized uptake value > 12 (P score 5) was ISUP ≥ 2 in all cases with 93% ISUP ≥ 3. The P score is easily calculated and improves accuracy for csPCa over both PI-RADS and PRIMARY scores. It should be considered when PSMA-PET is undertaken for diagnosis.

Interpretable machine learning model for early prediction of delirium in elderly patients following intensive care unit admission: a derivation and validation study.

Delirium is the most common neuropsychological complication among older adults admitted to the intensive care unit (ICU) and is often associated with a poor prognosis. This study aimed to construct and validate an interpretable machine learning (ML) for early delirium prediction in older ICU patients. This was a retrospective observational cohort study and patient data were extracted from the Medical Information Mart for Intensive Care-IV database. Feature variables associated with delirium, including predisposing factors, disease-related factors, and iatrogenic and environmental factors, were selected using least absolute shrinkage and selection operator regression, and prediction models were built using logistic regression, decision trees, support vector machines, extreme gradient boosting (XGBoost), k-nearest neighbors and naive Bayes methods. Multiple metrics were used for evaluation of performance of the models, including the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, recall, F1 score, calibration plot, and decision curve analysis. SHapley Additive exPlanations (SHAP) were used to improve the interpretability of the final model. Nine thousand seven hundred forty-eight adults aged 65 years or older were included for analysis. Twenty-six features were selected to construct ML prediction models. Among the models compared, the XGBoost model demonstrated the best performance including the highest AUC (0.836), accuracy (0.765), sensitivity (0.713), recall (0.713), and F1 score (0.725) in the training set. It also exhibited excellent discrimination with AUC of 0.810, good calibration, and had the highest net benefit in the validation cohort. The SHAP summary analysis showed that Glasgow Coma Scale, mechanical ventilation, and sedation were the top three risk features for outcome prediction. The SHAP dependency plot and SHAP force analysis interpreted the model at both the factor level and individual level, respectively. ML is a reliable tool for predicting the risk of critical delirium in elderly patients. By combining XGBoost and SHAP, it can provide clear explanations for personalized risk prediction and more intuitive understanding of the effect of key features in the model. The establishment of such a model would facilitate the early risk assessment and prompt intervention for delirium.

Post-selection Inference in Multiverse Analysis (PIMA): An Inferential Framework Based on the Sign Flipping Score Test.

When analyzing data, researchers make some choices that are either arbitrary, based on subjective beliefs about the data-generating process, or for which equally justifiable alternative choices could have been made. This wide range of data-analytic choices can be abused and has been one of the underlying causes of the replication crisis in several fields. Recently, the introduction of multiverse analysis provides researchers with a method to evaluate the stability of the results across reasonable choices that could be made when analyzing data. Multiverse analysis is confined to a descriptive role, lacking a proper and comprehensive inferential procedure. Recently, specification curve analysis adds an inferential procedure to multiverse analysis, but this approach is limited to simple cases related to the linear model, and only allows researchers to infer whether at least one specification rejects the null hypothesis, but not which specifications should be selected. In this paper, we present a Post-selection Inference approach to Multiverse Analysis (PIMA) which is a flexible and general inferential approach that considers for all possible models, i.e., the multiverse of reasonable analyses. The approach allows for a wide range of data specifications (i.e., preprocessing) and any generalized linear model; it allows testing the null hypothesis that a given predictor is not associated with the outcome, by combining information from all reasonable models of multiverse analysis, and provides strong control of the family-wise error rate allowing researchers to claim that the null hypothesis can be rejected for any specification that shows a significant effect. The inferential proposal is based on a conditional resampling procedure. We formally prove that the Type I error rate is controlled, and compute the statistical power of the test through a simulation study. Finally, we apply the PIMA procedure to the analysis of a real dataset on the self-reported hesitancy for the COronaVIrus Disease 2019 (COVID-19) vaccine before and after the 2020 lockdown in Italy. We conclude with practical recommendations to be considered when implementing the proposed procedure.

Subjective sleep more predictive of global cognitive function than objective sleep in older adults: A specification curve analysis

ObjectivesSleep is associated with cognitive function in older adults. In the current study, we examined this relationship from subjective and objective perspectives, and determined the robustness and dimensional specificity of the associations using a comprehensive modelling approach. MethodsMultiple dimensions of subjective (sleep quality and daytime sleepiness) and objective sleep (sleep stages, sleep parameters, sleep spindles, and slow oscillations), as well as subjectively reported and objectively measured cognitive function were collected from 55 older adults. Specification curve analysis was used to examine the robustness of correlations for the effects of sleep on cognitive function. ResultsRobust associations were found between sleep and objectively measured cognitive function, but not with subjective cognitive complaints. In addition, subjective sleep showed robust and consistent associations with global cognitive function, whereas objective sleep showed a more domain-specific association with episodic memory. Specifically, subjective sleep quality and daytime sleepiness correlated with global cognitive function, and objective sleep parameters correlated with episodic memory. ConclusionsOverall, associations between sleep and cognitive function in older adults depend on how they are measured and which specific dimensions of sleep and domains of cognitive function are considered. It highlights the importance of focusing on specific associations to ameliorate the detrimental effects of sleep disturbance on cognitive function in later life.

Application of interpretable machine learning algorithms to predict acute kidney injury in patients with cerebral infarction in ICU

BackgroundAcute kidney injury (AKI) is not only a complication but also a serious threat to patients with cerebral infarction (CI). This study aimed to explore the application of interpretable machine learning algorithms in predicting AKI in patients with cerebral infarction. MethodsThe study included 3920 patients with CI admitted to the Intensive Care Unit and Emergency Medicine of the Central Hospital of Lishui City, Zhejiang Province. Nine machine learning techniques, including XGBoost, logistics, LightGBM, random forest (RF), AdaBoost, GaussianNB (GNB), Multi-Layer Perceptron (MLP), support vector machine (SVM), and k-nearest neighbors (KNN) classification, were used to develop a predictive model for AKI in these patients. SHapley Additive exPlanations (SHAP) analysis provided visual explanations for each patient. Finally, model effectiveness was assessed using metrics such as average precision (AP), sensitivity, specificity, accuracy, F1 score, precision-recall (PR) curve, calibration plot, and decision curve analysis (DCA). ResultsThe XGBoost model performed better in the internal validation set and the external validation set, with an AUC of 0.940 and 0.887, respectively. The five most important variables in the model were, in order, glomerular filtration rate, low-density lipoprotein, total cholesterol, hemiplegia and serum kalium. ConclusionThis study demonstrates the potential of interpretable machine learning algorithms in predicting CI patients with AKI.

Synthetic minority over-sampling technique-enhanced machine learning models for predicting recurrence of postoperative chronic subdural hematoma.

Chronic subdural hematoma (CSDH) is a neurological condition with high recurrence rates, primarily observed in the elderly population. Although several risk factors have been identified, predicting CSDH recurrence remains a challenge. Given the potential of machine learning (ML) to extract meaningful insights from complex data sets, our study aims to develop and validate ML models capable of accurately predicting postoperative CSDH recurrence. Data from 447 CSDH patients treated with consecutive burr-hole irrigations at Wenzhou Medical University's First Affiliated Hospital (December 2014-April 2019) were studied. 312 patients formed the development cohort, while 135 comprised the test cohort. The Least Absolute Shrinkage and Selection Operator (LASSO) method was employed to select crucial features associated with recurrence. Eight machine learning algorithms were used to construct prediction models for hematoma recurrence, using demographic, laboratory, and radiological features. The Border-line Synthetic Minority Over-sampling Technique (SMOTE) was applied to address data imbalance, and Shapley Additive Explanation (SHAP) analysis was utilized to improve model visualization and interpretability. Model performance was assessed using metrics such as AUROC, sensitivity, specificity, F1 score, calibration plots, and decision curve analysis (DCA). Our optimized ML models exhibited prediction accuracies ranging from 61.0% to 86.2% for hematoma recurrence in the validation set. Notably, the Random Forest (RF) model surpassed other algorithms, achieving an accuracy of 86.2%. SHAP analysis confirmed these results, highlighting key clinical predictors for CSDH recurrence risk, including age, alanine aminotransferase level, fibrinogen level, thrombin time, and maximum hematoma diameter. The RF model yielded an accuracy of 92.6% with an AUC value of 0.834 in the test dataset. Our findings underscore the efficacy of machine learning algorithms, notably the integration of the RF model with SMOTE, in forecasting the recurrence of postoperative chronic subdural hematoma. Leveraging the RF model, we devised an online calculator that may serve as a pivotal instrument in tailoring therapeutic strategies and implementing timely preventive interventions for high-risk patients.

Exploring the multiverse of analysis options for the alcohol Stroop

The alcohol Stroop is a widely used task in addiction science to measure the theoretical concept of attentional bias (a selective attention to alcohol-related cues in the environment), which is thought to be associated with clinical outcomes (craving and consumption). However, recent research suggests findings from this task can be equivocal. This may be because the task has many different potential analysis pipelines, which increase researcher degrees of freedom when analysing data and reporting results. These analysis pipelines largely come from how outlying reaction times on the task are identified and handled (e.g. individual reaction times > 3 standard deviations from the mean are removed from the distribution; removal of all participant data if > 25% errors are made). We used specification curve analysis across two alcohol Stroop datasets using alcohol-related stimuli (one published and one novel) to examine the robustness of the alcohol Stroop effect to different analytical decisions. We used a prior review of this research area to identify 27 unique analysis pipelines. Across both data sets, the pattern of results was similar. The alcohol Stroop effect was present and largely robust to different analysis pipelines. Increased variability in the Stroop effect was observed when implementing outlier cut-offs for individual reaction times, rather than the removal of participants. Stricter outlier thresholds tended to reduce the size of the Stroop interference effect. These specification curve analyses are the first to examine the robustness of the alcohol Stroop to different analysis strategies, and we encourage researchers to adopt such analytical methods to increase confidence in their inferences across cognitive and addiction science.

Does Health Vulnerability Predict Voting for Right-Wing Populist Parties in Europe?

Abstract: Introduction: A recent article by Kavanagh et al. (2021) in American Political Science Review suggested that health vulnerability predicts voting patterns for the populist far-right. Aim: We sought to distinguish socioculturally conservative from antielite voting, expecting health vulnerability to predict only the former. Methods: We combined data from the European Social Survey (round 7) with the Chapel Hill Expert Survey (2014). We preregistered several other hypotheses regarding associations between health and voting and predetermined the smallest effect size of interest (SESOI: r = .05). Results: Health vulnerability did not predict socioculturally conservative voting, and the results for antielite voting were mixed. To investigate why our result differed from Kavanagh et al.’s, we reran their analysis employing covariate specification curve analyses. The statistical significance and the direction of the association between health vulnerability and voting depended on which covariates were included. Across 1,000 models with randomly drawn covariate specifications, 59.6% showed a positive, 29.8% a zero, and 10.6% a negative association between health vulnerability and voting for far-right populist parties. However, all effect sizes were more consistently smaller than our predetermined SESOI. Conclusions: Our study illustrates the necessity of causally justifying and preregistering all covariates and predetermining an SESOI.