Model-based Recursive Partitioning Research Articles

Heterogeneity in the association between social support and mental distress in old-age retirees – a computational approach using longitudinal cohort data

BackgroundMental distress among retirees and older people is a severe public health challenge, and information on new risk groups is needed. This study aims to identify subgroups of old-age retirees with varying associations between low social support and mental distress by applying model-based recursive partitioning (MOB).MethodsWe used the Helsinki Health Study follow-up survey data of old-age retired former municipal sector employees of the City of Helsinki, Finland. Phase 1 data were collected in 2000–2002, when all participants were employed, Phase 2 in 2007, Phase 3 in 2012, Phase 4 in 2017, and Phase 5 in 2022 (n = 4,466, 81% women). Social support and covariates were measured at each Phase 1–5 and the outcome, mental distress (Depression Anxiety Stress Scales [DASS-21]) was measured at a single occasion, during Phase 5. The three subscales and the common factor of general distress were analysed separately. An approach rooted in computational statistics was used to investigate risk factor heterogeneity in the association of low social support and mental distress. MOB combines decision trees with regression analysis to identify subgroups with the most significant heterogeneity among risk factors.ResultsMedian (IQR) general distress score from DASS-21 was 5.7 (3.0, 9.0), while Social Support Questionnaire number-score (SSQN) was 1.5 (1.15, 2.05). The primary effect modifier for the association between social support and general distress was education (p < 0.001). Those with high education had a different association of low social support and general distress than those with low or medium education. Additionally, the subgroup with low and medium education had a significant effect modification for age (p = 0.01). For the association between low social support and depressive symptoms, the moderating effect of education was dependent on gender, as men with medium–high education had the weakest association, while for women with medium–high education the association was strongest.ConclusionsOur results suggest that stratification by sociodemographic variables is justifiable when investigating risk factors of mental distress in old-age retirees. The incongruent association of low social support and depressive symptoms in men with medium–high education compared to women with medium–high education is a promising target for confirmatory research.

Methylation Analysis to Detect CIN3+ in High-Risk Human Papillomavirus-Positive Self-Samples From the Population-Based Cervical Cancer Screening Program

Since 2017, a self-sampling device has been introduced to the Dutch population-based screening program to enable higher participation rates. However, routine triage cytology cannot be performed on self-sampling material. Methylation is an alternative triage method that can be performed directly on DNA extracted from self-samples. Recently, we tested a set of 15 published cervical intraepithelial neoplasia grade 3 or worse (CIN3+)-specific methylation markers and found a panel of 3 markers with a sensitivity of 82% and a specificity of 74%. In this study, we determined the sensitivity and specificity of 2 commercial assays using quantitative methylation-specific PCR. DNA from the same cohort of high-risk human papillomavirus-positive self-sampled material obtained through the population-based screening program in the North of the Netherlands from women with CIN2 or less (<CIN3, n = 208) and women with CIN3+ (n = 96) was used. The QIAsure methylation test (consisting of 2 methylation markers) showed a sensitivity of 65% and a specificity of 72%, whereas the Gyntect assay (consisting of 6 methylation markers) showed a sensitivity of 59% and a specificity of 91% for CIN3+. When we compared all individual 23 methylation markers, receiver operating characteristic analysis showed an area under the curve of ≥0.7 for 11 of 23 markers (P < .001). By model-based recursive partitioning and robustness analysis, we found a panel with a better sensitivity compared with QIAsure and Gyntect (P < .001). This new panel, consisting of ITGA4, ASCL1, and FAM19A4, has a sensitivity of 84% and a specificity of 70%, similar to our previously identified panel (ANKRD18CP, LHX8, and EPB41L3). Thus, in addition to our previously identified panel, the combination of ITGA4, ASCL1, and FAM19A4 showed good diagnostic performance and potentially can replace cytology, thereby avoiding additional doctor visits for many women worldwide and reducing the time for decision making for referral to the gynecologist.

Detecting Differential Item Functioning in Multidimensional Graded Response Models With Recursive Partitioning.

Differential item functioning (DIF) is a common challenge when examining latent traits in large scale surveys. In recent work, methods from the field of machine learning such as model-based recursive partitioning have been proposed to identify subgroups with DIF when little theoretical guidance and many potential subgroups are available. On this basis, we propose and compare recursive partitioning techniques for detecting DIF with a focus on measurement models with multiple latent variables and ordinal response data. We implement tree-based approaches for identifying subgroups that contribute to DIF in multidimensional latent variable modeling and propose a robust, yet scalable extension, inspired by random forests. The proposed techniques are applied and compared with simulations. We show that the proposed methods are able to efficiently detect DIF and allow to extract decision rules that lead to subgroups with well fitting models.

Examining Financial Hardship and Caregiver Subgroups in Kinship Foster Placements: A Machine Learning Approach

Children placed with kinship foster parents can experience less disruption and stronger family ties than children in non-kinship placements. However, financial hardship can restrict kinship caregivers from taking in relatives’ children. This study investigated (1) kinship caregivers’ financial standing compared to a national subsample of caregivers and (2) whether certain factors moderate the likelihood that a kinship caregiver will be able to provide care for additional non-relative children without additional financial assistance from the Department of Social Services (DSS). This study utilized primary data from 345 relatives across North Carolina and nationally representative secondary data on 6394 individuals’ financial circumstances. One-sample t-tests and chi-square goodness-of-fit tests revealed that caregivers who participated in our study generally fared better financially than caregivers at the national level. Model-based recursive partitioning results showed that if an additional child is placed in the home, the caregiver’s perceived capacity to provide care without extra DSS support decreased by approximately 19%, with a greater decrease (35%) among a subgroup of caregivers with low financial well-being status. The heterogeneity in caregivers’ experiences, capacities, and financial needs buttresses the need for nuanced interventions and programs targeting these caregivers, enabling them to provide more stable care for children placed in their homes.

Heterogeneous Effects of Continuous Positive Airway Pressure in Non-Sleepy Obstructive Sleep Apnea on Cardiovascular Disease Outcomes: Post Hoc Machine Learning Analysis of the ISAACC Trial (ECSACT Study).

Rationale: Randomized controlled trials of continuous positive airway pressure (CPAP) therapy for cardiovascular disease (CVD) prevention among patients with obstructive sleep apnea (OSA) have been largely neutral. However, given that OSA is a heterogeneous disease, there may be unidentified subgroups demonstrating differential treatment effects. Objectives: We sought to apply a novel data-drive approach to identify nonsleepy OSA subgroups with heterogeneous effects of CPAP on CVD outcomes within the Impact of Sleep Apnea Syndrome in the Evolution of Acute Coronary Syndrome (ISAACC) study. Methods: Participants were randomly partitioned into two datasets. One for training (70%) our machine-learning model and a second (30%) for validation of significant findings. Model-based recursive partitioning was applied to identify subgroups with heterogeneous treatment effects. Survival analysis was conducted to compare treatment (CPAP vs. usual care [UC]) outcomes within subgroups. Results: A total of 1,224 nonsleepy OSA participants were included. Of 55 features entered into our model, only two appeared in the final model (i.e., average OSA event duration and hypercholesterolemia). Among participants at or below the model-derived average event duration threshold (19.5 s), CPAP was protective for a composite of CVD events (training hazard ratio [HR], 0.46; P = 0.002). For those with longer event duration (>19.5 s), an additional split occurred by hypercholesterolemia status. Among participants with longer event duration and hypercholesterolemia, CPAP resulted in more CVD events compared with UC (training HR, 2.24; P = 0.011). The point estimate for this harmful signal was also replicated in the testing dataset (HR, 1.83; P = 0.118). Conclusions: We discovered subgroups of nonsleepy OSA participants within the ISAACC study with heterogeneous effects of CPAP. Among the training dataset, those with longer OSA event duration and hypercholesterolemia had nearly 2.5 times more CVD events with CPAP compared with UC, whereas those with shorter OSA event duration had roughly half the rate of CVD events if randomized to CPAP.

Refining first-line treatment decision in RAS wildtype (RAS-WT) metastatic colorectal cancer (mCRC) by combining clinical biomarkers: Results of the randomized phase 3 trial FIRE-3 (AIO KRK0306).

13 Background: Optimal patient selection for first-line treatment targeting epithelial growth factor receptor (EGFR) in RAS-WT mCRC is based on primary tumor sidedness (PTS) with anti-EGFR being the preferred option for patients with left-sided mCRC (LC). Right-sided mCRCs (RC) are preferentially treated in combination with bevacizumab targeting vascular endothelial growth factor (VEGF). Here, improvement in patient selection was evaluated by combining clinical biomarkers beyond PTS using the randomized phase III trial FIRE-3. Methods: FIRE-3 evaluated first-line FOLFIRI (folinic acid, fluorouracil and irinotecan) plus cetuximab (FOLFIRI/Cet) versus FOLFIRI plus bevacizumab (FOLFIRI/Bev) in patients with RAS-WT mCRC. Besides PTS, further clinical biomarkers were evaluated in pairwise combinations using Cox regression models and model-based recursive partitioning with Weibull models to predict treatment benefit of either treatment arm regarding overall survival (OS): age, sex, liver-limited disease status (LLD) and baseline carcinoembryonic antigen serum level (CEA). The resulting P-values of second-order interactions were adjusted using Holm-Bonferroni correction. The model with the best test statistics and P-value was chosen for further evaluations. Results: In 400 patients with RAS-WT mCRC, a model combining PTS and LLD status best predicted treatment outcome of either treatment arm (c-index = 0.603, p=0.005). Here, a significant survival benefit of FOLFIRI/Cet over FOLFIRI/Bev was evident in patients with LC/non-LLD (HR 0.62, p=0.02) compared to LC/LLD (HR 0.83, p=0.40). In patients with RC, FOLFIRI/Bev was significantly associated with increased OS compared to FOLFIRI/Cet when patients suffered from non-LLD (HR 2.09, p=0.010). However, patients with RC/LLD rather had a benefit from FOLFIRI/Cet compared to FOLFIRI/Bev (HR 0.59, p=0.218). Conclusions: Combining clinical biomarkers PTS and LLD status might improve optimal patient selection for targeted first-line treatment in RAS-WT mCRC. Validation in further data sets is warranted. Clinical trial information: NCT00433927 .

On recursive partitioning to refine coordinate rotation in Eddy covariance applications

While the use of three-dimensional sonic anemometers for eddy covariance is well established, there remain questions about its deployment and consequent data analysis. Convention has largely accepted that coordinates must be rotated so that the corrected transverse and vertical mean velocities become zero. In non-ideal terrain, a planar fit coordinate rotation is widely accepted. However, if the coordinate system changes through time, such as in a rapidly growing maize canopy, or is affected by the local topography, finding temporally and directionally stable regions for a reference point is crucial for flux estimation. A model-based recursive partitioning (MOB) algorithm is proposed to answer this problem. The algorithm creates individual wind sectors and accounts for temporal or spatially dependent variables that may affect planar fits. A cross-validation fitting procedure of the MOB algorithm has proved crucial in identifying variables that influence the choice among different parametric models, such as day of year, solar altitude, and wind direction. Momentum exchange is the most sensitive to alternative coordinate rotation systems, especially in stable conditions. The MOB methodology would be most useful during these conditions. The analyses presented here show that in quantifying scalar fluxes during the day, especially in the case of sensible heat flux and by extension to evapotranspiration and carbon dioxide exchange, there is little need for coordinate rotation if the sensors were aligned with the vertical parallel to gravity.

A model for the effective design of customized recycling programs through audience segmentation and modularity approaches

One-size-fits-all approaches to encouraging household recycling behavior remain ineffective in densely populated urban areas due to their failure to address the heterogeneity in household needs/motivations for recycling. In contrast, extensive service customization escalates costs, requiring finding ways to use available resources effectively. We aim to develop a model to resolve this dilemma based on empirical research on 526 households sampled from a densely populated district of İzmir, Turkey. We applied a model-based recursive partitioning method to identify recycling behavior segments with their critical determinant factors. We then conducted a focus group study to investigate strategies for exploiting commonalities and considering differences across segments regarding their determinant factors. The decision tree analysis revealed the existence of 12 segments differentiated by the determinant factors of task knowledge, perceived convenience, moral norms, social norms, how to dispose of waste, and marital status. The focus group study suggested a range of informative, operational, and social intervention methods that address these factors in pairs or individually. Drawing on these findings, we develop a model that guides the design of recycling programs in which customized intervention methods can influence multiple segments simultaneously. Our empirical analysis shows that the methods elevating ‘perceived convenience’ and ‘task knowledge’ even slightly above the moderate level would increase the likelihood of recycling for more than half of households. Our model further suggests customized intervention methods which consider the segments' sizes, their current recycling likelihoods, and the potential improvements in these. This study contributes mainly by developing a model for designing customized recycling programs that maximize household contributions with the available budget/resources. Public managers can use our model to design effective recycling programs considering budget limitations.

Detecting heterogeneity in the causal direction of dependence: A model-based recursive partitioning approach.

Methods of causal discovery and direction of dependence to evaluate causal properties of variable relations have experienced rapid development. The majority of causal discovery methods, however, relies on the assumption of causal effect homogeneity, that is, the identified causal structure is expected to hold for the entire population. Because causal mechanisms can vary across subpopulations, we propose combining methods of model-based recursive partitioning and non-Gaussian causal discovery to identify such subpopulations. The resulting algorithm can discover subpopulations with potentially varying magnitude and causal direction of effects under mild parameter inequality assumptions. Feasibility conditions are described and results from synthetic data experiments are presented suggesting that large effects and large sample sizes are beneficial for detecting causally competing subgroups with acceptable statistical performance. In a real-world data example, the extraction of meaningful subgroups that differ in the causal mechanism underlying the development of numerical cognition is illustrated. Potential extensions and recommendations for best practice applications are discussed.

Treatment effect of remdesivir on the mortality of hospitalised COVID-19 patients in Switzerland across different patient groups: a tree-based model analysis.

Remdesivir has shown benefits against COVID-19. However, it remains unclear whether, to what extent, and among whom remdesivir can reduce COVID-19-related mortality. We explored whether the treatment response to remdesivir differed by patient characteristics. We analysed data collected from a hospital surveillance study conducted in 21 referral hospitals in Switzerland between 2020 and 2022. We applied model-based recursive partitioning to group patients by the association between treatment levels and mortality. We included either treatment (levels: none, remdesivir within 7 days of symptom onset, remdesivir after 7 days, or another treatment), age and sex, or treatment only as regression variables. Candidate partitioning variables included a range of risk factors and comorbidities (and age and sex unless included in regression). We repeated the analyses using local centring to correct the results for the propensity to receive treatment. Overall (n = 21,790 patients), remdesivir within 7 days was associated with increased mortality (adjusted hazard ratios 1.28-1.54 versus no treatment). The CURB-65 score caused the most instability in the regression parameters of the model. When adjusted for age and sex, patients receiving remdesivir within 7 days of onset had higher mortality than those not treated in all identified eight patient groups. When age and sex were included as partitioning variables instead, the number of groups increased to 19-20; in five to six of those branches, mortality was lower among patients who received early remdesivir. Factors determining the groups where remdesivir was potentially beneficial included the presence of oncological comorbidities, male sex, and high age. Some subgroups of patients, such as individuals with oncological comorbidities or elderly males, may benefit from remdesivir.

Identification of a methylation panel as an alternative triage to detect CIN3+ in hrHPV-positive self-samples from the population-based cervical cancer screening programme

BackgroundThe Dutch population-based cervical cancer screening programme (PBS) consists of primary high-risk human papilloma virus (hrHPV) testing with cytology as triage test. In addition to cervical scraping by a general practitioner (GP), women are offered self-sampling to increase participation. Because cytological examination on self-sampled material is not feasible, collection of cervical samples from hrHPV-positive women by a GP is required. This study aims to design a methylation marker panel to detect CIN3 or worse (CIN3+) in hrHPV-positive self-samples from the Dutch PBS as an alternative triage test for cytology.MethodsFifteen individual host DNA methylation markers with high sensitivity and specificity for CIN3+ were selected from literature and analysed using quantitative methylation-specific PCR (QMSP) on DNA from hrHPV-positive self-samples from 208 women with CIN2 or less (< CIN2) and 96 women with CIN3+. Diagnostic performance was determined by area under the curve (AUC) of receiver operating characteristic (ROC) analysis. Self-samples were divided into a train and test set. Hierarchical clustering analysis to identify input methylation markers, followed by model-based recursive partitioning and robustness analysis to construct a predictive model, was applied to design the best marker panel.ResultsQMSP analysis of the 15 individual methylation markers showed discriminative DNA methylation levels between < CIN2 and CIN3+ for all markers (p < 0.05). The diagnostic performance analysis for CIN3+ showed an AUC of ≥ 0.7 (p < 0.001) for nine markers. Hierarchical clustering analysis resulted in seven clusters with methylation markers with similar methylation patterns (Spearman correlation> 0.5). Decision tree modeling revealed the best and most robust panel to contain ANKRD18CP, LHX8 and EPB41L3 with an AUC of 0.83 in the training set and 0.84 in the test set. Sensitivity to detect CIN3+ was 82% in the training set and 84% in the test set, with a specificity of 74% and 71%, respectively. Furthermore, all cancer cases (n = 5) were identified.ConclusionThe combination of ANKRD18CP, LHX8 and EPB41L3 revealed good diagnostic performance in real-life self-sampled material. This panel shows clinical applicability to replace cytology in women using self-sampling in the Dutch PBS programme and avoids the extra GP visit after a hrHPV-positive self-sampling test.

Predicting heterogeneous treatment effects of an Internet-based depression intervention for patients with chronic back pain: Secondary analysis of two randomized controlled trials

BackgroundDepression is highly prevalent among individuals with chronic back pain. Internet-based interventions can be effective in treating and preventing depression in this patient group, but it is unclear who benefits most from this intervention format. MethodIn an analysis of two randomized trials (N = 504), we explored ways to predict heterogeneous treatment effects of an Internet-based depression intervention for patients with chronic back pain. Univariate treatment-moderator interactions were explored in a first step. Multilevel model-based recursive partitioning was then applied to develop a decision tree model predicting individualized treatment benefits. ResultsThe average effect on depressive symptoms was d = −0.43 (95 % CI: −0.68 to –0.17; 9 weeks; PHQ-9). Using univariate models, only back pain medication intake was detected as an effect moderator, predicting higher effects. More complex interactions were found using recursive partitioning, resulting in a final decision tree with six terminal nodes. The model explained a large amount of variation (bootstrap-bias-corrected R2 = 45 %), with predicted subgroup-conditional effects ranging from di = 0.24 to −1.31. External validation in a pilot trial among patients on sick leave (N = 76; R2 = 33 %) pointed to the transportability of the model. ConclusionsThe studied intervention is effective in reducing depressive symptoms, but not among all chronic back pain patients. Predictions of the multivariate tree learning model suggest a pattern in which patients with moderate depression and relatively low pain self-efficacy benefit most, while no benefits arise when patients' self-efficacy is already high. If corroborated in further studies, the developed tree algorithm could serve as a practical decision-making tool.

Rasch-tree 방법을 이용한 PISA 2015 과학검사의 차별기능문항 탐색

The purpose of this study is to introduce the Rasch-tree, which is Rasch model-based recursive partitioning, as a new method for differential item functioning(DIF) detection and to examine its usefulness by exploring DIF in PISA 2015 Science test data with the Rasch-tree. This study used 16 items binary response of PISA 2015 Science test and 41 variables related to the student, parent, family, teacher, science, and ICT. The results of this study showed that there are four subgroups splitted by the three variables, ‘Enjoyment of Science(JOYSCICE)’, ‘gender’, and ‘Child’s past science activities(PRESUPP)’, and the two DIF items were detected. These items had large differences in the estimate item difficulty parameters between these four subgroups, especially the group of students with high level of JOYSCIE and the group of female students with relatively low level of JOYSCIE. For these results, it was confirmed that the Rasch-tree could be effective for exploring DIF between latent classes by considering simultaneously multiple trait variables.

Do Multiple Sex/Gender Dimensions Play a Role in the Association of Green Space and Self-Rated Health? Model-Based Recursive Partitioning Results from the KORA INGER Study.

Exposure to green space has a positive impact on health. Whether sex/gender modifies the green space-health association has so far only been studied through the use of a binary sex/gender category; however, sex/gender should be considered more comprehensively as a multidimensional concept based on theoretical approaches. We therefore explored whether sex/gender, operationalized through multiple sex/gender- and intersectionality-related covariates, modifies the green space-self-rated health association. We collected data from participants involved in the German KORA study (Cooperative Health Research in the Region of Augsburg) in 2019. Self-rated health was assessed as a one-question item. The availability of green spaces was measured subjectively as well as objectively. The multiple sex/gender- and intersectionality-related covariates were measured via self-assessment. To analyze the data, we used model-based recursive partitioning, a decision tree method that can handle complex data, considering both multiple covariates and their possible interactions. We showed that none of the covariates operationalizing an individual sex/gender self-concept led to subgroups with heterogeneous effects in the model-based tree analyses; however, we found effect heterogeneity based on covariates representing structural aspects from an intersectionality perspective, although they did not show the intersectional structuring of sex/gender dimensions. In one identified subgroup, those with a lower education level or a feeling of discrimination based on social position showed a positive green space-self-rated health association, while participants with a higher education level or no feeling of discrimination based on social position had a high level of self-rated health regardless of the availability of green spaces. Model-based recursive partitioning has the potential to detect subgroups exhibiting different exposure-outcome associations, with the possibility of integrating multiple sex/gender- and intersectionality-related covariates as potential effect modifiers. A comprehensive assessment of the relevance of sex/gender showed effect heterogeneity based on covariates representing structural aspects from an intersectionality perspective.

Intensive care management for high-risk veterans in a patient-centered medical home – do some veterans benefit more than others?

BackgroundPrimary care intensive management programs utilize interdisciplinary care teams to comprehensively meet the complex care needs of patients at high risk for hospitalization. The mixed evidence on the effectiveness of these programs focuses on average treatment effects that may mask heterogeneous treatment effects (HTEs) among subgroups of patients. We test for HTEs by patients’ demographic, economic, and social characteristics. MethodsRetrospective analysis of a VA randomized quality improvement trial. 3995 primary care patients at high risk for hospitalization were randomized to primary care intensive management (n = 1761) or usual primary care (n = 1731). We estimated HTEs on ED and hospital utilization one year after randomization using model-based recursive partitioning and a pre-versus post-with control group framework. Splitting variables included administratively collected demographic characteristics, travel distance, copay exemption, risk score for future hospitalizations, history of hospital discharge against medical advice, homelessness, and multiple residence ZIP codes. ResultsThere were no average or heterogeneous treatment effects of intensive management one year after enrollment. The recursive partitioning algorithm identified variation in effects by risk score, homelessness, and whether the patient had multiple residences in a year. Within each distinct subgroup, the effect of intensive management was not statistically significant. ConclusionsPrimary care intensive management did not affect acute care use of high-risk patients on average or differentially for patients defined by various demographic, economic, and social characteristics. ImplicationsReducing acute care use for high-risk patients is complex, and more work is required to identify patients positioned to benefit from intensive management programs.

Employing model-based recursive partitioning to identify neurophysiological subtypes of depression and corresponding symptom characteristics

Employing model-based recursive partitioning to identify neurophysiological subtypes of depression and corresponding symptom characteristics

Modeling preference heterogeneity using model-based decision trees

This article investigates the usage of a general model-based recursive partitioning algorithm to model preference heterogeneity. We use the algorithm to grow a decision tree based on statistical tests of the stability of individuals’ preference parameters. In particular, we used a Mixed Logit (MIXL) model with alternative-specific attributes at the end leaves of the tree while using individual characteristics as partition variables. This configuration allows us to search for instabilities of the taste parameters across individuals’ characteristics. We conduct a simulation study to investigate the algorithm’s ability to recover different data generating processes with structural breaks in the taste parameters. The results show that the algorithm can correctly recover diverse tree-like data generating processes. Additionally, we applied the algorithm to stated choice data of the preferences for the environmental impact of (hypothetical) energy generation plans in Chile. The results show that the model-based decision tree fits the data better than MIXL in terms of information criteria. Moreover, we show that the derived tree structure depends on the assumptions on the parameters’ distributions. Additionally, we compare the model-based decision tree model with Latent Class (LC) models with and without within-class heterogeneity. Finally, we show that the recursive partitioning algorithm can inform the selection of variables to be included in the LC allocation models.

Development and validation of a prognostic model for the early identification of COVID-19 patients at risk of developing common long COVID symptoms

BackgroundThe coronavirus disease 2019 (COVID-19) pandemic demands reliable prognostic models for estimating the risk of long COVID. We developed and validated a prediction model to estimate the probability of known common long COVID symptoms at least 60 days after acute COVID-19.MethodsThe prognostic model was built based on data from a multicentre prospective Swiss cohort study. Included were adult patients diagnosed with COVID-19 between February and December 2020 and treated as outpatients, at ward or intensive/intermediate care unit. Perceived long-term health impairments, including reduced exercise tolerance/reduced resilience, shortness of breath and/or tiredness (REST), were assessed after a follow-up time between 60 and 425 days. The data set was split into a derivation and a geographical validation cohort. Predictors were selected out of twelve candidate predictors based on three methods, namely the augmented backward elimination (ABE) method, the adaptive best-subset selection (ABESS) method and model-based recursive partitioning (MBRP) approach. Model performance was assessed with the scaled Brier score, concordance c statistic and calibration plot. The final prognostic model was determined based on best model performance.ResultsIn total, 2799 patients were included in the analysis, of which 1588 patients were in the derivation cohort and 1211 patients in the validation cohort. The REST prevalence was similar between the cohorts with 21.6% (n = 343) in the derivation cohort and 22.1% (n = 268) in the validation cohort. The same predictors were selected with the ABE and ABESS approach. The final prognostic model was based on the ABE and ABESS selected predictors. The corresponding scaled Brier score in the validation cohort was 18.74%, model discrimination was 0.78 (95% CI: 0.75 to 0.81), calibration slope was 0.92 (95% CI: 0.78 to 1.06) and calibration intercept was −0.06 (95% CI: −0.22 to 0.09).ConclusionThe proposed model was validated to identify COVID-19-infected patients at high risk for REST symptoms. Before implementing the prognostic model in daily clinical practice, the conduct of an impact study is recommended.

Model-based Recursive Partitioning to Estimate Unfair Health Inequalities in the United Kingdom Household Longitudinal Study

We measure unfair health inequality in the UK using a novel data-driven empirical approach. We explain health variability as the result of circumstances beyond individual control and health-related behaviours. We do this using model-based recursive partitioning, a supervised machine learning algorithm. Unlike usual tree-based algorithms, model-based recursive partitioning does identify social groups with different expected levels of health but also unveils the heterogeneity of the relationship linking behaviors and health outcomes across groups. The empirical application is conducted using the UK Household Longitudinal Study. We show that unfair inequality is a substantial fraction of the total explained health variability. This finding holds no matter which exact definition of fairness is adopted: using both the fairness gap and direct unfairness measures, each evaluated at different reference values for circumstances or effort.

Analysis of overcorrection to be included for planning clear aligner therapy: a retrospective study.

To provide clinical information on overcorrection to be included in the initial digital setup to make clear aligner therapy (CAT) more efficient. Prescription data for 150 patients (80 women and 70 men; mean age 33.7 ± 12.7 years) treated successfully with CAT (F22 Aligners, Sweden & Martina, Due Carrare, Italy) and requiring only a single, minimal finishing phase were acquired retrospectively. The inclusion criteria were Class I dental malocclusion with only minimal crowding (≤3 mm), 12-20 aligner steps per arch, no use of auxiliaries or interarch elastics, and rotations ≤25° for round-shaped teeth. The prescribed and corrective movements to be achieved in the main and finishing treatment phases, respectively, were quantified by the dedicated clear aligner setup software. The magnitudes of inclination (buccal-lingual crown tipping), angulation (mesial-distal crown tipping), rotation, intrusion, and extrusion were extracted and analyzed by tooth type, maxilla and mandible, and both arches. Descriptive statistics, that is, mean, standard deviation, and percentage, were calculated for each movement investigated. Classification and regression trees (CART) were generated using the model-based recursive partitioning approach, and the corrective movements were correlated with respect to both the amount of the movements prescribed and the tooth type. Statistical significance was set at 5%. Inclination and rotation required the greatest correction, whereas angulation, intrusion, and extrusion required only minimal correction. Expressed as a percentage of prescribed movement, mean corrective movements were 20.5% for inclination, 14.5% angulation, 28.4% rotation, 11.7% extrusion, and 22% intrusion. According to CART, all corrective movements except extrusion depended on both tooth type and the magnitude of prescribed movement. To achieve more efficient CAT, approximately 20% overcorrection should be added to the initial planning phase when planning challenging movements such as inclination and rotation.