Recursive Partitioning Method Research Articles

Exploration of Fingerprints and Data Mining-based Prediction of Some Bioactive Compounds from Allium sativum as Histone Deacetylase 9 (HDAC9) Inhibitors.

Histone deacetylase 9 (HDAC9) is an important member of the class IIa family of histone deacetylases. It is well established that over-expression of HDAC9 causes various types of cancers including gastric cancer, breast cancer, ovarian cancer, liver cancer, lung cancer, lymphoblastic leukaemia, etc. The important role of HDAC9 is also recognized in the development of bone, cardiac muscles, and innate immunity. Thus, it will be beneficial to find out the important structural attributes of HDAC9 inhibitors for developing selective HDAC9 inhibitors with higher potency. The classification QSAR-based methods namely Bayesian classification and recursive partitioning method were applied to a dataset consisting of HADC9 inhibitors. The structural features strongly suggested that sulphur-containing compounds can be a good choice for HDAC9 inhibition. For this reason, these models were applied further to screen some natural compounds from Allium sativum. The screened compounds were further accessed for the ADME properties and docked in the homology-modelled structure of HDAC9 in order to find important amino acids for the interaction. The best-docked compound was considered for molecular dynamics (MD) simulation study. The classification models have identified good and bad fingerprints for HDAC9 inhibition. The screened compounds like ajoene, 1,2 vinyl dithiine, diallyl disulphide and diallyl trisulphide had been identified as compounds having potent HDAC9 inhibitory activity. The results from ADME and molecular docking study of these compounds show the binding interaction inside the active site of the HDAC9. The best-docked compound ajoene shows satisfactory results in terms of different validation parameters of MD simulation study. This in-silico modelling study has identified the natural potential lead (s) from Allium sativum. Specifically, the ajoene with the best in-silico features can be considered for further in-vitro and in-vivo investigation to establish as potential HDAC9 inhibitors.

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.

Prognostic Role of Multiparameter Flow Cytometry-Based Measurable Residual Disease Assessment in Acute Myeloid Leukemia Patients with FMS3-like Tyrosine Kinase-3 Internal Tandem Duplication (FLT3-ITD)

Introduction Measurable residual disease (MRD) monitoring is predictive in acute myeloid leukemia (AML). Assessment of FMS3-like tyrosine kinase-3 in-frame internal tandem duplications (FLT3-ITD) is usually performed at diagnosis by polymerase chain reaction (PCR). Due to its relative lower sensitivity of 2%, FLT3-ITD PCR is not routinely used during response assessment. Next-generation sequencing is similarly limited by technical difficulties in capturing tandem duplications using the short base pair-based sequencing method. Long base pair-based sequencing has been reported, but its use is limited by financial restrictions. Multiparameter flow cytometry (MFC) can be a useful tool for MRD monitoring in this AML subtype until such time that molecular techniques for detecting FLT3-ITD MRD are optimized. Patients and methods The study evaluated the outcomes of FLT3-ITD mutated AML patients diagnosed and treated from 2018 to 2022 at Princess Margaret Cancer Centre. We compared outcomes according to MFC-MRD post-induction and FLT3-ITD allele frequency (AF) status at diagnosis. MRD cut-off was 0.1%. Data were locked as of June 30, 2023. Clinical outcomes evaluated include overall survival (OS) and relapse-free survival (RFS). The cumulative incidence of relapse (CIR) and non-relapse mortality (NRM) were calculated considering competing risk. The Kaplan-Meier method using a log-rank test and a multivariate Cox proportional hazard model was used for analyses, while the Gray test and Fine-Grey model were used for uni- and multivariate analysis for CIR and NRM. Results A total of 111 patients with a mean age of 63.5 years were included, of whom 90 received treatment. Secondary AML accounted for 12.7% of patients. Risk stratification according to European LeukemiaNet (ELN) 2022 was favorable in 2 (1.8%), intermediate in 67 (60.4%), and adverse in 42 patients (37.8%). Nucleophosmin 1 (NPM1) co-mutation was observed in 55 patients (49.5%). Seventy-nine patients (87.8%) could be assessed for overall response, including 69 (76.7%) who achieved complete remission (CR) or CR with incomplete count recovery (CRi). Of these, 54 achieved first CR/CRi (CR1) with 1 induction cycle. MFC-MRD data were available in 61 patients, of whom 44 (72.1%) were MRD negative, while 17 (27.9%) were MRD positive. With a median follow-up of 437 days, 50 patients (45%) were still alive. Median OS and RFS were 3.42 years and 1.05 years, respectively. Among patients who achieved CR1, post-induction MFC-MRD positivity correlated with an inferior OS (HR 2.35 [1.06-5.26], p=0.037) and a trend for a shorter RFS (HR 2.08 [0.99-4.35], p=0.052). We examined the impact of FLT3-ITD AF at diagnosis on long-term outcomes. By applying a binary recursive partitioning method, the cut-off of FLT3-ITD AF with the best risk stratification power for RFS, was defined at 54.6%. The group with a higher FLT3-ITD AF showed inferior OS (HR 1.86 [1.01-3.44], p=0.047) and RFS (HR 1.91 [1.09-3.33], p=0.023). Taking together FLT3-ITD AF at diagnosis and MFC-MRD status at CR1, patients with low FLT3-ITD AF and negative MRD had the highest OS rate of 86.2% at 12 months (p=0.023), and the highest RFS at 72.4% (p=0.096), while the corresponding values for low FLT3-ITD AF/positive MRD patients were 87.5% and 50%, respectively. In contrast, those with high FLT3-ITD AF/negative MRD had an OS of 72.7% and a RFS of 45.5%, while those with high FLT3-ITD AF/positive MRD showed the lowest OS (16.7%) and the shortest RFS (16.7%). There was no statistical difference in CIR and NRM among groups. Multivariate analysis with stepwise selection was performed, considering age at diagnosis, ELN 2022 risk, MFC-MRD at CR1, FLT3-ITD AF at diagnosis, cytogenetics, and NPM1 co-mutation. Predictive factors for OS were MFC-MRD post-induction (HR 2.49 [1.11-5.61], p=0.027) and FLT3-ITD AF (HR 2.29 [1.03-5.11], p=0.043). For RFS, age at diagnosis (HR 1.03 [1.00-1.06], p=0.038) and FLT3-ITD AF (HR 2.40 [1.15-5.01], p=0.019) were predictive, while MFC-MRD was not significant (HR 1.59 [0.72-3.53], p=0.25). Conclusion Our data demonstrate that MFC-based MRD assessment is feasible in AML with FLT3-ITD. Although better outcomes are expected in patients with a lower FLT3-ITD AF, patients who failed to achieve MRD negativity at CR1 showed inferior outcomes. The presence of both poor risk factors, a high AF of FLT3-ITD at diagnosis and MRD positivity at CR1, correlated with the worst treatment outcomes.

Real World Experience of FLT3-Inhibitor Post-Transplant Maintenance with Sorafenib Demonstrating Superior Overall- and Relapse Free Survival Following Allogeneic Hematopoietic Stem Cell Transplantation in FLT3-ITD Mutated Acute Myeloid Leukemia

Introduction Acute myeloid leukemia (AML) with FLT3-internal tandem duplication (ITD) is known to have a significantly high risk of relapse even with allogeneic stem cell transplantation (HCT), for which post-transplant maintenance therapy with FLT3 inhibitors (FLT3i) has been applied. Although there is a randomized trial comparing Sorafenib maintenance vs no treatment, real world experience is still limited. Despite the increased use of FTL3i maintenance in clinical practice, the optimal dose and duration of post-transplant Sorafenib maintenance remains uncertain. Accordingly, we performed a retrospective chart review to analyze the survival benefit with Sorafenib maintenance, dose tolerability and the impact of FLT3-ITD allele frequency (AF) at initial diagnosis of AML on treatment outcomes in the context of post-HCT Sorafenib maintenance. Patients and method We conducted a retrospective study including 69 patients who received an allogeneic HCT from 2019 to 2023 for AML with FLT3-ITD, of which 24 patients received Sorafenib as post-HCT maintenance therapy. Primary endpoint was relapse-free survival (RFS). The use of Sorafenib maintenance treatment was treated as a time-dependent covariate and Mantel-Byar test (MBT) was conducted to compare outcomes between Sorafenib maintenance group vs others, which will avoid immortal bias. FLT3-ITD allele frequency (AF) was assessed by molecular PCR test, and the cutoff was determined using a binary recursive partitioning method with AF of 54.6% as a cutoff. Results Out of 69 patients, 61 (88%) were in CR1 vs 8 (12%) in CR2 prior to HCT. Sorafenib was started in 24 patients (35%) at a median of 95 days post-HCT with a dose of 200mg every other day (n=2), 200mg once daily (n=17) or 200mg twice daily (n=5). The most common maintenance dose was 200mg once daily (n=9), followed by 200mg twice daily (n=6), 200mg every other day (n=5), and 400mg twice daily (n=1). The median Sorafenib maintenance duration was 15 months (1-53 months) at the last follow-up. There were no relapses in the patients who completed 2 years of Sorafenib maintenance. Sorafenib was discontinued in 6 patients, due to relapse (n=2), death of unrelated cause (n=1), GVHD (n=1), pneumonitis (n=1) and recurrent hypoglycemia (n=1). Adverse events requiring Sorafenib dose reduction included, GI intolerance (n=7), skin rash (n=7), thrombocytopenia (n=4), transaminitis (n=3), neuropathy (n=3), GVHD (n=2) and pneumonitis (n=2). The median follow-up duration for all the patients was 27 months (5-51) following HCT, or 13 months (1.3-48) following Sorafenib maintenance. Using Kaplan-Meier test, the 2-year RFS in patients on Sorafenib maintenance was 84.5% vs 40.5% for no Sorafenib (p=0.00049). The 2-year OS was 88.2% vs 53.9% (p=0.0043), respectively. The CIR was 9.8% vs 39.9% (p=0.00634), while the NRM was 5.6% vs 19.7% (p =0.122), respectively. In the MBT, Sorafenib maintenance was found to reduce the mortality risk by 62.5% (HR 0.375 (0.109-1.29], p=0.120) and reduced the relapse/mortality risk by 71.2% (HR 0.288 [0.084-0.984], p=0.047). With respect to FLT3-ITD AF, the group with higher AF showed extremely higher HR for RFS compared to those with lower FLT-ITD (HR 29.37 [3.007-286.8], p=0.0036). When evaluating RFS and OS with combined risk incorporating the use of Sorafenib and FLT3-ITD AF, we have 4 subgroups (figure 1). In patients with a lower AF, the RFS was 87.7% vs 56.8% in the Sorafenib (n=20), vs no Sorafenib group (n=31), respectively. In patients with a higher AF, the RFS was 66.7% vs 0.0% in the Sorafenib (n=3), vs no Sorafenib group (n=8; p=0.0009), respectively. In a multivariate analysis, we confirmed the 2-year RFS for patients who had a high FLT3-ITD and received Sorafenib maintenance was statistically significant when compared to no Sorafenib maintenance. Conclusion Sorafenib maintenance post-transplant in FLT3-ITD mutated AML patients significantly improves the OS and RFS. Sorafenib maintenance is tolerable with 200mg twice daily or lower in most patients. Reduced doses of Sorafenib improve tolerability, with no significant impact on increased risk of relapse. Higher AF group of FLT3-ITD at initial diagnosis defined by ≥54.6% using PCR, was associated with significantly increased risk of relapse even after HCT. Sorafenib maintenance seems to provide survival benefit in terms of RFS, particularly in the patients with higher AF FLT3-ITD.

A Comparative Analysis of Data Mining Methods and Hierarchical Linear Modeling Using PISA 2018 Data

A Comparative Analysis of Data Mining Methods and Hierarchical Linear Modeling Using PISA 2018 Data

Single-nucleotide polymorphisms of nucleotide excision repair genes and their association with overall survival in North Indian lung cancer patients treated with platinum-based doublet chemotherapy.

The aim of the study is to understand the association of Nucleotide excision repair (NER) inter-genetic polymorphic combinations with overall survival (OS) in lung cancer as well as its histological subtypes in the North Indian population. Genotyping was performed using polymerase chain reaction-restriction fragment length polymorphism. For survival analysis, the univariate Kaplan-Meier and multivariate Cox regression model were applied. Recursive partitioning method, survival analysis tree was applied to study unfavorable genotypic combinations in NER single-nucleotide polymorphisms. Combinatorial studies suggested no association between polymorphic combinations of NER genes and OS in lung cancer patients. When stratified as per lung cancer histological subtypes, adenocarcinomas patients with XPG 670 and XPC 499 polymorphism a significant increase in OS in combined heterozygous and mutant genotype with a lower hazardous ratio (Hazard Ratio1 (HR) = 0.20; P = 0.004). Small-cell lung carcinoma (SCLC) patients with XPF 11985A>G and XPD Arg156Arg polymorphism showed a 4-fold hazard ratio among heterozygous genotype (HR1 = 4.84; P = 0.007) where no significant results are obtained in patients suffering from squamous cell carcinoma histological subtypes. STREE showed XPG Asp1104His (W), XPD Lys751Gln (H + M), XPF Arg415Gln (H + M) genotype was associated with a lower hazard ratio (P = 0.0007) showing survival of 11.6 months when compared with the reference (median survival time = 3.52). It can be concluded that SCLC patients with polymorphic combinations of the NER pathway were associated with a higher risk of mortality. STREE depicted the association of NER polymorphic combinations with a lower hazard ratio predicting them to be a good prognostic factor for lung cancer.

A hybrid approach for the analysis of complex categorical data structures: assessment of latent distance learning perception in higher education

A long tradition of analysing ordinal response data deals with parametric models, which started with the seminal approach of cumulative models. When data are collected by means of Likert scale survey questions in which several scored items measure one or more latent traits, one of the sore topics is how to deal with the ordered categories. A stacked ensemble (or hybrid) model is introduced in the proposal to tackle the limitations of summing up the items. In particular, multiple items responses are synthesised into a single meta-item, defined via a joint data reduction approach; the meta-item is then modelled according to regression approaches for ordered polytomous variables accounting for potential scaling effects. Finally, a recursive partitioning method yielding trees provides automatic variable selection. The performance of the method is evaluated empirically by using a survey on Distance Learning perception.

Admission Triage With Pain, Inspiratory Effort, Cough Score can Predict Critical Care Utilization and Length of Stay in Isolated Chest Wall Injury

Damage to the thoracic cage is common in the injured patient, both when the injuries are confined to this single cavity and as part of the overall injury burden of a polytraumatized patient. In a subset of these patients, the severity of injury to the intrathoracic viscera is either underappreciated at admission or blossom over the following 48-72h. The ability to promptly identify these patients and abrogate complications therefore requires triage of such at-risk patients to close monitoring in a critical care environment. At our institution, this triage hinges on the Pain, Inspiratory effort, Cough (PIC) score, which generates a composite unitless score from a nomogram which aggregates several variables-patient-reported Pain visual analog scale, Incentive spirometry effort, and the perceived adequacy of Cough. We thus sought to audit PIC's discriminant power in predicting intensive care unit (ICU) need. This retrospective cohort study was performed at an urban, academic, level 1 trauma center. All isolated chest wall injuries (excluded any Abbreviated Injury Score >2 in head or abdomen) from January 2020 to June 2021 were identified in the local trauma registry. The electronic medical record was queried for standard demographics, admission PIC score, postadmission destination, ICU and hospital length of stay (LOS), and any unplanned admissions to the ICU. Chi-squared tests were used to determine differences between PIC score outcomes and the recursive partitioning method correlated admission PIC score to ICU LOS. Two hundred and thirty six isolated chest wall injury patients were identified, of whom 194 were included in the final analysis. The median age was 60 (interquartile range [IQR] 50-74) years, 63.1% were male, and the median (IQR) number of rib fractures was 3.0 (2.0-5.0). A cutoff PIC score of 7 or lower was associated with ICU admission (odds ratio [OR] 95% CI: 8.19 [3.39-22.55], P<0.001 with a PPV=41.4%, NPV=91%), and with ICU admission for greater than 48h [OR (95% CI): 26.86 (5.5-43.96), P<0.001, with a PPV=25.9%, NPV=98.7%] but not anatomic injury severity score, hospital LOS or ICU, or the requirement for mechanical ventilation. The association between PIC score 7 or below and the presence of bilateral fractures, flail chest, or sternal fracture did not meet statistical significance. The accurate cut point of the PIC score to predict ICU admission over 48h in our retrospective cohort was calculated as PIC≤7 for P=0.013 and PIC≤6 for P=0.001. Patients with isolated chest wall injuries require effective reproducible triage for ICU-level care. The PIC score appears to be a moderate discriminator of critical care need, per se, as judged by our recorded complication rate requiring critical care intervention. This vigilance may pay dividends in early detection and abrogation of respiratory failure emergencies. Furthermore, PIC score delineation for ICU need appears to be appropriate at 7 or less; this threshold can be used during admission triage to guide care.

Multiple Bonus–Malus Scale Models for Insureds of Different Sizes

How to consider the a priori risks in experience-rating models has been questioned in the actuarial community for a long time. Classic past-claim-rating models, such as the Buhlmann–Straub credibility model, normalize the past experience of each insured before applying claim penalties. On the other hand, classic Bonus–Malus Scales (BMS) models generate the same surcharges and the same discounts for all insureds because the transition rules within the class system do not depend on the a priori risk. Despite the quality of prediction of the BMS models, this experience-rating model could appear unfair to many insureds and regulators because it does not recognize the initial risk of the insured. In this paper, we propose the creation of different BMSs for each type of insured using recursive partitioning methods. We apply this approach to real data for the farm insurance product of a major Canadian insurance company with widely varying sizes of insureds. Because the a priori risk can change over time, a study of the possible transitions between different BMS models is also performed.

Alternative expectation approaches for expectation-maximization missing data imputations in cox regression

Missing data is common in survival analysis. It is either removed or imputed using various methods. Expectation-maximization (EM) imputation is a popular method in Cox regression studies. This paper investigated the effect of different regression methods on Cox regression modeling within the framework of EM. A stratified Cox regression model was derived from a dataset of categorical and numerical variables. Missing data were imputed using the EM framework with five machine learning algorithms and then were compared to the full model. The results show that the recursive partition and regression tree (RPART) method performed better than others. However, all regression methods performed poorly in categorical covariate imputation. R code is available online.

Determinants of left ventricular function improvement for cardiac resynchronization therapy candidates.

AimsA waiting period of more than 3 months is recommended for patients before undergoing cardiac resynchronization therapy (CRT). However, due to an anticipated high mortality rate, early implementation of CRT might be beneficial for some patients. We aimed to evaluate the rate and the probability of left ventricular (LV) function improvement and their predictors in patients with heart failure (HF) with indications for CRT.Methods and resultsFrom March 2011 to February 2014, a total of 5625 hospitalized patients for acute HF were consecutively enrolled in 10 tertiary hospitals. Among them, we analysed 1792 patients (mean age 63.96 ± 15.42 years, female 63.1%) with left ventricular ejection fraction (LVEF) ≤ 35% at the baseline echocardiography and divided them into three groups: 144 with left bundle branch block (LBBB), 136 with wide QRS complexes without LBBB, and 1512 not having these findings (control). We compared and analysed these three groups for improvement of LV function at follow‐up echocardiography. In patients who met CRT indications (patients with LBBB or wide QRS complexes without LBBB), logistic regression was performed to identify risk factors for no improvement of LV. No improvement of LV was defined as LVEF ≤ 35% at follow‐up echocardiography or the composite adverse outcomes: death, heart transplantation, extracorporeal membrane oxygenation, or use of a ventricular assist device before follow‐up echocardiography. A classification tree was established using the binary recursive partitioning method to predict the outcome of patients who met CRT indications. In a median follow‐up of 11 months, LVEF improvement was observed in 24.3%, 15.4%, and 40.5% of patients with LBBB, wide QRS complexes without LBBB, and control, respectively. Patients meeting CRT indications had higher 3 month mortality rates than the control (24.6% vs. 17.7%, P = 0.002). Multivariable logistic regression analysis revealed that large LV end‐systolic dimension [odds ratio (OR) 1.10, 95% confidence interval (CI) 1.05–1.15, P < 0.001], low LVEF (OR 0.92, 95% CI 0.87–0.98, P = 0.006), diabetes requiring insulin (OR 6.49, 95% CI 2.53–19.33, P < 0.001), and suboptimal medical therapy (OR 6.85, 95% CI 3.21–15.87, P < 0.001) were significant factors predictive of no improvement. A decision tree analysis was consistent with these results.ConclusionsPatients with CRT indications had higher mortality during their follow‐up compared with control. LV function improvement was rare in this population, especially when they had some risk factors. These results suggest that the uniform waiting period before CRT implantation could be reconsidered and individualized.

BCR–ABL1 transcript doubling time as a predictor for treatment‐free remission failure after imatinib discontinuation in chronic myeloid leukaemia in chronic phase

The doubling time (DT) of the BCR-ABL1 quantitative polymerase chain reaction (qPCR) transcript level reflects the re-growing fraction of leukaemic cells after discontinuation of tyrosine kinase inhibitor (TKI). The present study analyzed monthly DT within six months after imatinib discontinuation in 131 patients. Monthly DT was calculated as x = ln(2)/K, where x is the DT and K is the fold BCR-ABL1 change from the previous value divided by the number of days between each measurement. The optimal DT value was determined as 12·75 days at two months using a recursive partitioning method. The patients were stratified into three groups: the high-risk group (DT<12·75 days but >0, with rapidly proliferating chronic myeloid leukaemia (CML) cells; n = 26) showed the lowest molecular relapse-free survival (mRFS) of 7·7% at 12 months, compared to 53·6% in the intermediate-risk group (DT≥12·75 days, with slowly proliferating CML cells; n = 16) or 90·0% in the low-risk group (DT≤0, i.e., without proliferating CML cells; n = 71; P < 0·001). Monthly assessment of DT helps identify high-risk patients for treatment-free remission failure with an imminent risk of molecular recurrence, and to define low-risk patients who can be spared the frequent monitoring of monthly molecular tests.

A new multi-level algorithm for balanced partition problem on large scale directed graphs

Graph partition is a classical combinatorial optimization and graph theory problem, and it has a lot of applications, such as scientific computing, VLSI design and clustering etc. In this paper, we study the partition problem on large scale directed graphs under a new objective function, a new instance of graph partition problem. We firstly propose the modeling of this problem, then design an algorithm based on multi-level strategy and recursive partition method, and finally do a lot of simulation experiments. The experimental results verify the stability of our algorithm and show that our algorithm has the same good performance as METIS. In addition, our algorithm is better than METIS on unbalanced ratio.

Data-Driven Subgroup Identification in Confirmatory Clinical Trials.

Data-driven subgroup analysis plays an important role in clinical trials. This paper focuses on practical considerations in post-hoc subgroup investigations in the context of confirmatory clinical trials. The analysis is aimed at assessing the heterogeneity of treatment effects across the trial population and identifying patient subgroups with enhanced treatment benefit. The subgroups are defined using baseline patient characteristics, including demographic and clinical factors. Much progress has been made in the development of reliable statistical methods for subgroup investigation, including methods based on global models and recursive partitioning. The paper provides a review of principled approaches to data-driven subgroup identification and illustrates subgroup analysis strategies using a family of recursive partitioning methods known as the SIDES (subgroup identification based on differential effect search) methods. These methods are applied to a Phase III trial in patients with metastatic colorectal cancer. The paper discusses key considerations in subgroup exploration, including the role of covariate adjustment, subgroup analysis at early decision points and interpretation of subgroup search results in trials with a positive overall effect.

Pre-Transplant Pulmonary Function Test Parameters of FEF50% (Forced expiratory flow at 50% of forced vital capacity) and RV (residual volume) in Combination Can Predict the Risk for Development of Lung Gvhd after Allogeneic Transplantation

BackgroundChronic graft vs host disease (GVHD) of the lung is a significant post-transplant complication that impacts patients' quality of life and overall survival. It manifests as bronchiolitis obliterans syndrome (BOS), an irreversible progressive airflow obstruction. Risk factors identified include busulfan, peripheral blood stem cell source, post-transplant viral infections, history of lung disease, and acute GVHD. Diagnosis involves changes in pulmonary function tests (PFTs), ruling out infection and presence of chronic GVHD in other organ systems. Without routine monitoring patients can present with moderate to severe symptoms. The predictive value of pre-transplant PFTs has not been clearly elucidated. In this retrospective study, we examined the role of pre-transplant PFT parameters with respect to the development of lung GVHD.MethodsPatients receiving allogeneic HCT at Princess Margaret Cancer Centre between 2004 to 2013 were included in the analysis. Pre-HCT PFT results were routinely obtained within 30 days prior to transplant. PFT parameters included % predicted FEV1, DLCO, TLC, RV, RV/TLC, FRC, FVC, FEV1/FVC, FEF50% and FEF75%. After transplant, PFT was repeated around day 120, then annually, or when clinically indicated.Our definition of lung GVHD required at least the first two of the 2005 NIH consensus criteria for the diagnosis of chronic lung GVHD: 1) FEV1/forced vital capacity ratio <0.7 and FEV1<75% predicted, 2) evidence of airway obstruction on high-resolution computed tomography or a residual volume on PFTs >120%, 3) absence of infection in the respiratory tract, and 4) histologic diagnosis of BOS or at least 1 other manifestation of chronic GVHD in an additional organ system.The primary endpoint was cumulative incidence (CI) of lung GVHD at 3 years. The cut-off values in each PFT parameter with respect to development of lung GVHD were calculated using binary recursive partitioning (rpart) method provided by rpart package using R. Pearson's correlation test was used to look for relationship among the PFT parameters. The parameters with cut-offs identified by rpart were analyzed by factor analysis with Varimax rotation method.For multivariable analysis, Fine-Gray proportional hazard regression with competing risks was applied. The clinical variables included in the model were busulfan, acute GVHD, and stem cell source (peripheral vs bone marrow). All statistical tests were performed using EZR and SPSS.ResultsA total of 605 patients with available PFTs were reviewed. The CI of lung GVHD at 3 years was 10.6% (7.9 - 13.7) with a median onset at 12.6 months (range of 1.7 - 55.8).Univariate analysis of each PFT parameter as a continuous variable did not show an impact on development of GVHD. With recursive partitioning, cut-offs were identified for DLCO (68%), FEF50% (39%), FRC (130%) and RV (93%).In Pearson's correlation analysis, DLCO and FEF50% did not have a significant association with the other PFT parameters selected from rpart, while RV was significantly associated with TGV (r= 0.65). In the factor analysis, based on the factor loading, FEF50% and RV were selected for multivariate analysis.Multivariate analysis showed that FEF50% < 39% and RV ≥ 93% were independent risk factors of lung GVHD development. The patients with pre-transplant FEF50 ≥ 39% had significantly less risk compared to those with < 39% (p=0.025; HR 0.389), and RV ≥ 93% showed a 2 fold higher risk than those < 93% (p=0.036, HR 2.63).A risk score model was created using the cut-offs from FEF50% and RV. For FEF50%, score 0 was assigned to ≥ 39% while score 1 was < 39%. For RV, score 0 was assigned to < 93%, while score 1 was ≥ 93%. A risk score was summed from the two parameters. Risk scores of 2 were classified as high risk, scores of 1 intermediate risk and score of 0 low risk. The 3 year CIs of lung GVHD were then compared according to the new risk score. The high risk group had an incidence of 33.5% (11.8 - 57.0%; n=28, 4.6%), intermediate risk 11.4% (8.3 - 15.2%; n=445, 73.6%) and low risk 3.0% (0.8 - 8.2%; n=132, 21.8%; p = 0.003; Figure).Conclusions1) Pre-transplant PFT parameters can predict the risk of lung GVHD development.2) FEF50% and RV were predictive of lung GVHD development independent from the other clinical risk factors acute GVHD, busulfan and stem cell source which were not confirmed in our cohort.3) A risk score model using pre-transplant FEF50% and RV may help in prediction of lung GVHD. [Display omitted] DisclosuresKim:Paladin: Consultancy; Novartis: Consultancy, Honoraria, Research Funding; BMS: Consultancy, Honoraria, Research Funding; Pfizer: Consultancy.

Recursive Partitioning Methods to Examine the Effects of Physical Activity on Physical and Mental Health: A Case Study at One Public University in the United States

The purpose of this study was to examine the effects of physical activity (PA) on physical and mental health within an adult sample, while controlling for demographic characteristics. Data were analyzed from the 2018–2019 Exercise is Medicine® On Campus (EIM-OC) assessment of students, faculty, and staff at one public university in the United States. Participants completed a survey including a demographic questionnaire, the 12-Item Short Form Health Survey (SF-12), and the International Physical Activity Questionnaire (IPAQ). Classification and Regression Trees, a supervised learning technique based upon recursive partitioning, were utilized to explore the relationships between PA and health. Variable-importance was assessed via the permutation method and confirmed with ensemble learning methods. Results indicated that PA, especially in leisure, was predictive of both physical and mental health. Further, leisure PA was the single most important variable contributing to mental health. Greater levels of physical health were observed in adults achieving at least 672 MET minutes of leisure PA per week, while greater levels of mental health were observed in adults achieving at least 950 MET minutes of leisure PA per week. Despite current recommendations for PA (irrespective of domain), more PA (especially in leisure) improves physical and mental health.

Prognostic classification in acute exacerbation of idiopathic pulmonary fibrosis: a multicentre retrospective cohort study

Acute exacerbation (AE) in idiopathic pulmonary fibrosis (IPF) is a major prognostic determinant. However, evidence for its prognostic strength is mainly based on the results of small cohort studies with statistical limitations. This retrospective study, which included 108 patients with a first episode of AE-IPF, aimed to identify prognostic factors and to develop prognostic classification models. Multivariate Cox regression analysis revealed that a lower percent-predicted forced vital capacity within 12 months before AE onset (baseline %FVC) and a lower PaO2/FiO2 ratio at AE onset were independent mortality predictors. If the value of each predictor was lower than the cutoff determined by receiver-operating characteristic analysis, 1 point was assigned. Classification of patients into mild, moderate, and severe groups based on total score showed post-AE 90-day cumulative survival rates of 83.3%, 66.2%, and 22.2%, respectively (model 1: C-index 0.702). Moreover, a decision tree-based model was created with the recursive partitioning method using baseline %FVC and PaO2/FiO2 ratio at AE onset from among multivariable; accordingly, patients were classified into 3 groups with post-AE 90-day cumulative survival rates of 84.1%, 64.3%, and 24.0%, respectively (model 2: C-index 0.735). These models can guide clinicians in determining therapeutic strategies and help design future studies on AE-IPF.

Identification of subgroup effect with an individual participant data meta-analysis of randomised controlled trials of three different types of therapist-delivered care in low back pain

BackgroundProven treatments for low back pain, at best, only provide modest overall benefits. Matching people to treatments that are likely to be most effective for them may improve clinical outcomes and makes better use of health care resources.MethodsWe conducted an individual participant data meta-analysis of randomised controlled trials of three types of therapist delivered interventions for low back pain (active physical, passive physical and psychological treatments). We applied two statistical methods (recursive partitioning and adaptive risk group refinement) to identify potential subgroups who might gain greater benefits from different treatments from our individual participant data meta-analysis.ResultsWe pooled data from 19 randomised controlled trials, totalling 9328 participants. There were 5349 (57%) females with similar ratios of females in control and intervention arms. The average age was 49 years (standard deviation, SD, 14).Participants with greater psychological distress and physical disability gained most benefit in improving on the mental component scale (MCS) of SF-12/36 from passive physical treatment than non-active usual care (treatment effects, 4.3; 95% confidence interval, CI, 3.39 to 5.15). Recursive partitioning method found that participants with worse disability at baseline gained most benefit in improving the disability (Roland Morris Disability Questionnaire) outcome from psychological treatment than non-active usual care (treatment effects, 1.7; 95% CI, 1.1 to 2.31). Adaptive risk group refinement did not find any subgroup that would gain much treatment effect between psychological and non-active usual care. Neither statistical method identified any subgroups who would gain an additional benefit from active physical treatment compared to non-active usual care.ConclusionsOur methodological approaches worked well and may have applicability in other clinical areas. Passive physical treatments were most likely to help people who were younger with higher levels of disability and low levels of psychological distress. Psychological treatments were more likely to help those with severe disability. Despite this, the clinical importance of identifying these subgroups is limited. The sizes of sub-groups more likely to benefit and the additional effect sizes observed are small. Our analyses provide no evidence to support the use of sub-grouping for people with low back pain.

Tree-structured scale effects in binary and ordinal regression

In binary and ordinal regression one can distinguish between a location component and a scaling component. While the former determines the location within the range of the response categories, the scaling indicates variance heterogeneity. In particular since it has been demonstrated that misleading effects can occur if one ignores the presence of a scaling component, it is important to account for potential scaling effects in the regression model, which is not possible in available recursive partitioning methods. The proposed recursive partitioning method yields two trees: one for the location and one for the scaling. They show in a simple interpretable way how variables interact to determine the binary or ordinal response. The developed algorithm controls for the global significance level and automatically selects the variables that have an impact on the response. The modeling approach is illustrated by several real-world applications.

Measuring everyday creativity: A Rasch model analysis of the Biographical Inventory of Creative Behaviors (BICB) scale

Research on everyday creativity—the “little c” creative activities people do in their everyday lives—commonly uses self-report scales to assess people’s engagement in different activities. The present research presents a detailed psychometric analysis of the Biographical Inventory of Creative Behaviors (BICB), a 34-item yes/no checklist of common creative activities that has become one of the most popular self-report measures of everyday creative behaviors. Based on a sample of 2,359 adults, the reliability, dimensionality, item fit, item difficulty, and test information were evaluated from a Rasch model perspective. Overall, the BICB shows good evidence for score reliability and appears essentially unidimensional; a small cluster of misfitting and locally dependent items were flagged for impairing unidimensionality. The items’ difficulty level was generally moderate and suitable for the scale’s intended populations and purposes. Differential item functioning (DIF) based on gender and age, estimated via Rasch tree recursive partitioning methods, found notable gender-based DIF (generally reflecting culturally gendered qualities of some creative activities) but little age-based DIF. Taken together, the BICB has many psychometric strengths. Some opportunities for future scale refinement are discussed.