Poly-bagging predictors for classification modelling for credit scoring

Credit scoring modelling comprises one of the leading formal tools for supporting the granting of credit. Its core objective consists of the generation of a score by means of which potential clients can be listed in the order of the probability of default. A critical factor is whether a credit scoring model is accurate enough in order to provide correct classification of the client as a good or bad payer. In this context the concept of bootstraping aggregating (bagging) arises. The basic idea is to generate multiple classifiers by obtaining the predicted values from the fitted models to several replicated datasets and then combining them into a single predictive classification in order to improve the classification accuracy. In this paper we propose a new bagging-type variant procedure, which we call poly-bagging, consisting of combining predictors over a succession of resamplings. The study is derived by credit scoring modelling. The proposed poly-bagging procedure was applied to some different artificial datasets and to a real granting of credit dataset up to three successions of resamplings. We observed better classification accuracy for the two-bagged and the three-bagged models for all considered setups. These results lead to a strong indication that the poly-bagging approach may promote improvement on the modelling performance measures, while keeping a flexible and straightforward bagging-type structure easy to implement. (C) 2011 Elsevier Ltd. All rights reserved.

Predictors of health status do not change over three-year periods and exacerbation makes difference in chronic obstructive pulmonary disease

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

The variability and predictors of quality of AIDS care services in Brazil

Background: Since establishing universal free access to antiretroviral therapy in 1996, the Brazilian Health System has increased the number of centers providing HIV/AIDS outpatient care from 33 to 540. There had been no formal monitoring of the quality of these services until a survey of 336 AIDS health centers across 7 Brazilian states was undertaken in 2002. Managers of the services were asked to assess their clinics according to parameters of service inputs and service delivery processes. This report analyzes the survey results and identifies predictors of the overall quality of service delivery.Methods: The survey involved completion of a multiple-choice questionnaire comprising 107 parameters of service inputs and processes of delivering care, with responses assessed according to their likely impact on service quality using a 3-point scale. K-means clustering was used to group these services according to their scored responses. Logistic regression analysis was performed to identify predictors of high service quality.Results: The questionnaire was completed by 95.8% (322) of the managers of the sites surveyed. Most sites scored about 50% of the benchmark expectation. K-means clustering analysis identified four quality levels within which services could be grouped: 76 services (24%) were classed as level 1 (best), 53 (16%) as level 2 (medium), 113 (35%) as level 3 (poor), and 80 (25%) as level 4 (very poor). Parameters of service delivery processes were more important than those relating to service inputs for determining the quality classification. Predictors of quality services included larger care sites, specialization for HIV/AIDS, and location within large municipalities.Conclusion: The survey demonstrated highly variable levels of HIV/AIDS service quality across the sites. Many sites were found to have deficiencies in the processes of service delivery processes that could benefit from quality improvement initiatives. These findings could have implications for how HIV/AIDS services are planned in Brazil to achieve quality standards, such as for where service sites should be located, their size and staffing requirements. A set of service delivery indicators has been identified that could be used for routine monitoring of HIV/AIDS service delivery for HIV/AIDS in Brazil (and potentially in other similar settings).

BODE Index and GOLD Staging as Predictors of 1-Year Exacerbation Risk in Chronic Obstructive Pulmonary Disease

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Gender differences in predictors of health status in patients with COPD

Objective: To evaluate the health status (HS) of COPD patients and to identify the main predictors of HS in these patients according to gender. Methods: The study included 90 COPD patients (60 males and 30 females; mean age = 64 +/- 9 years) with a wide range of airway obstruction disorders (mean FEV(1) = 56 +/- 19% of predicted). The men were individually matched to the women by % of predicted FEV(1) (ratio: 2:1). The patients were assessed regarding body composition; six-minute walk distance; perception of dyspnea using the Modified Medical Research Council Dyspnea Scale; Saint George's Respiratory Questionnaire (SGRQ); Charlson comorbidity index; and the multidimensional Body mass index, airway Obstruction, Dyspnea, and Exercise capacity (BODE) index. Multiple linear regression analysis was performed to identify the predictors of HS by gender. Results: Impairment of HS was greater among the women than among the men for SGRQ total score and for all SGRQ domains (total: 51 +/- 18% vs. 38 19%; p = 0.002; symptoms: 61 +/- 22% vs. 42 +/- 21%; p < 0.001; activity: 62 +/- 18% vs. 49 +/- 21%; p = 0.004; and impact: 41 +/- 19% vs. 27 +/- 18%; p = 0.001). The multiple linear regression showed that age and perception of dyspnea correlated with SGRQ total score for both genders (males, r(2) = 0.42; females, r(2) = 0.70; p < 0.05). Conclusions: Our results showed an association between gender and HS in COPD patients. Age and dyspnea are determinants of HS in both genders.

Serum Metalloproteinases 2 and 9 as Predictors of Gait Status, Pressure Ulcer and Mortality after Hip Fracture

Introduction: The aim of this study is to evaluate the serum activity of metalloproteinases (MMPs) -2 and -9 as predictors of pressure ulcer (PU), gait status and mortality 6 months after hip fracture. Methods: Eighty-seven patients over the age of 65 admitted to the orthopedic unit from January to December 2010 with hip fracture were prospectively evaluated. Upon admission, patient demographic information, including age, gender and concomitant diseases, was recorded. Blood samples were taken for analysis of MMP -2 and -9 activity by gel zymography and for biochemical examination within the first 72 hours of the patient's admission, after clinical stabilization. The fracture pattern (neck, trochanteric or subtrochanteric), time from admission to surgery, surgery duration and length of hospital stay were also recorded. Results: Two patients were excluded due to the presence of pathological fractures (related to cancer), and three patients were excluded due to the presence of PU before admission. Eighty-two patients, with a mean age of 80.4 ± 7.3 years, were included in the analysis. Among these patients, 75.6% were female, 59.8% had PU, and 13.4% died 6 months after hip fracture. All patients underwent hip fracture repair. In a univariate analysis, there were no differences in serum MMP activity between hip fracture patients with or without PU. In addition, the multiple logistic regression analysis models, which were adjusted by age, gender, length of hospital stay and C-reactive protein, showed that the pro-MMP-9 complexed with neutrophil gelatinase-associated lipocalin form (130 kDa) was associated with gait status recovery 6 months after hip fracture. Conclusions: In conclusion, serum pro-MMP-9 is a predictor of gait status recovery 6 months after hip fracture. © 2013 Gumieiro et al.

Frequency and clinical predictors of coronary artery disease in chronic renal failure renal transplant candidates

Background/aims: Cardiovascular diseases are major causes of mortality in chronic renal failure patients before and after renal transplantation. Among them, coronary disease presents a particular risk; however, risk predictors have been used to diagnose coronary heart disease. This study evaluated the frequency and importance of clinical predictors of coronary artery disease in chronic renal failure patients undergoing dialysis who were renal transplant candidates, and assessed a previously developed scoring system. Methods: Coronary angiographies conducted between March 2008 and April 2013 from 99 candidates for renal transplantation from two transplant centers in Sao Paulo state were analyzed for associations between significant coronary artery diseases (>= 70% stenosis in one or more epicardial coronary arteries or >= 50% in the left main coronary artery) and clinical parameters. Results: Univariate logistic regression analysis identified diabetes, angina, and/or previous infarction, clinical peripheral arterial disease and dyslipidemia as predictors of coronary artery disease. Multiple logistic regression analysis identified only diabetes and angina and/or previous infarction as independent predictors. Conclusion: The results corroborate previous studies demonstrating the importance of these factors when selecting patients for coronary angiography in clinical pretransplant evaluation.

Clinical Predictors Associated With Duration of Repetitive Transcranial Magnetic Stimulation Treatment for Remission in Bipolar Depression A Naturalistic Study

Repetitive transcranial magnetic stimulation (rTMS) has been widely tested and shown to be effective for unipolar depression. Although it has also been investigated for bipolar depression (BD), there are only few rTMS studies with BD. Here, we investigated 56 patients with BD who received rTMS treatment until remission (defined as Hamilton Depression Rating Scores <= 7). We used simple and multiple logistic regressions to identify clinical and demographic predictors associated with duration of treatment (defined as <15 vs. >15 rTMS sessions). Age, refractoriness, number of prior depressive episodes, and severe depression at baseline were associated with a longer rTMS treatment. In the multivariate analysis, refractoriness (likelihood ratio (LR) = 4.33; p < 0.01) and baseline severity (LR = 0.18, p < 0.01) remained significant predictors. Our preliminary study showed that, in remitted patients, refractoriness and severity of index episode are associated with the need of a longer rTMS treatment; providing preliminary evidence of important factors associated with rTMS parameters adjustment.

Frequency and predictors of symptomatic intracranial hemorrhage after intravenous thrombolysis for acute ischemic stroke in a Brazilian public hospital

OBJECTIVE: Scarce data are available on the occurrence of symptomatic intracranial hemorrhage related to intravenous thrombolysis for acute stroke in South America. We aimed to address the frequency and clinical predictors of symptomatic intracranial hemorrhage after stroke thrombolysis at our tertiary emergency unit in Brazil. METHOD: We reviewed the clinical and radiological data of 117 consecutive acute ischemic stroke patients treated with intravenous thrombolysis in our hospital between May 2001 and April 2010. We compared our results with those of the Safe Implementation of Thrombolysis in Stroke registry. Univariate and multiple regression analyses were performed to identify factors associated with symptomatic intracranial transformation. RESULTS: In total, 113 cases from the initial sample were analyzed. The median National Institutes of Health Stroke Scale score was 16 (interquartile range: 10-20). The median onset-to-treatment time was 188 minutes (interquartile range: 155-227). There were seven symptomatic intracranial hemorrhages (6.2%; Safe Implementation of Thrombolysis in Stroke registry: 4.9%; p = 0.505). In the univariate analysis, current statin treatment and elevated National Institute of Health Stroke Scale scores were related to symptomatic intracranial hemorrhage. After the multivariate analysis, current statin treatment was the only factor independently associated with symptomatic intracranial hemorrhage. CONCLUSIONS: In this series of Brazilian patients with severe strokes treated with intravenous thrombolysis in a public university hospital at a late treatment window, we found no increase in the rate of symptomatic intracranial hemorrhage. Additional studies are necessary to clarify the possible association between statins and the risk of symptomatic intracranial hemorrhage after stroke thrombolysis.

Clinical predictors of long-term outcome in obsessive-compulsive disorder

Background: The purpose of this study was to investigate demographic and clinical factors associated with the long-term outcome of obsessive-compulsive disorder (OCD). Methods: A hundred ninety-six previously untreated patients with DSM-IV criteria OCD completed a 12-week randomized open trial of group cognitive-behavioral therapy (GCBT) or fluoxetine, followed by 21 months of individualized, uncontrolled treatment, according to international guidelines for OCD treatment. OCD severity was assessed using the Yale–Brown Obsessive-Compulsive Scale (Y-BOCS) at different times over the follow-up period. Demographics and several clinical variables were assessed at baseline. Results: Fifty percent of subjects improved at least 35% from baseline, and 21.3% responded fully (final Y-BOCS score < or = 8). Worse prognosis was associated with earlier age at onset of OCD (P = 0.045), longer duration of illness (P = 0.001) presence of at least one comorbid psychiatric disorder (P = 0.001), comorbidity with a mood disorder (P = 0.002), higher baseline Beck-Depression scores (P = 0.011), positive family history of tics (P = 0.008), and positive family history of anxiety disorders (P = 0.008). Type of initial treatment was not associated with long-term outcome. After correction for multiple testing, the presence of at least one comorbid disorder, the presence of a depressive disorder, and duration of OCD remained significant. Conclusions: Patients under cognitive-behavioral or pharmacological treatment improved continuously in the long run, regardless of initial treatment modality or degree of early response, suggesting that OCD patients benefit from continuous treatment. Psychiatric comorbidity, especially depressive disorders, may impair the long-term outcome of OCD patients.

The impact of polymorphisms in P-gp, DNA repair and folic acid metabolism genes in newly diagnosed multiple myeloma patients treated with thalidomide plus dexamethasone, with or without bortezomib

The principle aim of this study was to investigate biological predictors of response and resistance to multiple myeloma treatment. Two hypothesis had been proposed as responsible of responsiveness: SNPs in DNA repair and Folate pathway, and P-gp dependent efflux. As a first objective, panel of SNPs in DNA repair and Folate pathway genes, were analyzed. It was a retrospective study in a group of 454, previously untreated, MM patients enrolled in a randomized phase III open-label study. Results show that some SNPs in Folate pathway are correlated with response to MM treatment. MTR genotype was associated with favorable response in the overall population of MM patients. However, this relation, disappear after adjustment for treatment response. When poor responder includes very good partial response, partial response and stable/progressive disease MTFHR rs1801131 genotype was associated with poor response to therapy. This relation - unlike in MTR – was still significant after adjustment for treatment response. Identification of this genetic variant in MM patients could be used as an independent prognostic factor for therapeutic outcome in the clinical practice. In the second objective, basic disposition characteristics of bortezomib was investigated. We demonstrated that bortezomib is a P-gp substrate in a bi-directional transport study. We obtain apparent permeability rate values that together with solubility values can have a crucial implication in better understanding of bortezomib pharmacokinetics with respect to the importance of membrane transporters. Subsequently, in view of the importance of P-gp for bortezomib responsiveness a panel of SNPs in ABCB1 gene - coding for P-gp - were analyzed. In particular we analyzed five SNPs, none of them however correlated with treatment responsiveness. However, we found a significant association between ABCB1 variants and cytogenetic abnormalities. In particular, deletion of chromosome 17 and t(4;14) translocation were present in patients harboring rs60023214 and rs2038502 variants respectively.

Corpus callosum index and long-term disability in multiple sclerosis patients

Prediction of long-term disability in patients with multiple sclerosis (MS) is essential. Magnetic resonance imaging (MRI) measurement of brain volume may be of predictive value but sophisticated MRI techniques are often inaccessible in clinical practice. The corpus callosum index (CCI) is a normalized measurement that reflects changes of brain volume. We investigated medical records and 533 MRI scans at diagnosis and during clinical follow-up of 169 MS patients (mean age 42 +/- 11 years, 86% relapsing-remitting MS, time since first relapse 11 +/- 9 years). CCI at diagnosis was 0.345 +/- 0.04 and correlated with duration of disease (p = 0.002; r = -0.234) and expanded disability status scale (EDSS) score at diagnosis (r = -0.428; p < 0.001). Linear regression analyses identified age, duration of disease, relapse rate and EDSS at diagnosis as independent predictors for disability after mean of 7.1 years (Nagelkerkes' R:0.56). Annual CCI decrease was 0.01 +/- 0.02 (annual tissue loss: 1.3%). In secondary progressive MS patients, CCI decrease was double compared to that in relapsing-remitting MS patients (p = 0.04). There was a trend of greater CCI decrease in untreated patients compared to those who received disease modifying drugs (p = 0.2). CCI is an easy to use MRI marker for estimating brain atrophy in patients with MS. Brain atrophy as measured with CCI was associated with disability progression but it was not an independent predictor of long-term disability.

Prevalence, predictors, and consequences of long-term refusal of antipsychotic treatment in first-episode psychosis

Studies investigating medication adherence in psychosis are limited by the need of a certain degree of medication adherence and the inclusion of mostly multiple-episode patients. By contrast, noninformed consent, epidemiological studies in first-episode psychosis (FEP) allow the assessment of an important subgroup of patients who persistently refuse antipsychotic medication and thereby never receive an adequate antipsychotic trial. The present study aims to assess the prevalence and predictors of such a "medication refusal" subgroup and its association with illness outcome.

Predictors and overestimation of recalled mobile phone use among children and adolescents

A growing body of literature addresses possible health effects of mobile phone use in children and adolescents by relying on the study participants' retrospective reconstruction of mobile phone use. In this study, we used data from the international case-control study CEFALO to compare self-reported with objectively operator-recorded mobile phone use. The aim of the study was to assess predictors of level of mobile phone use as well as factors that are associated with overestimating own mobile phone use. For cumulative number and duration of calls as well as for time since first subscription we calculated the ratio of self-reported to operator-recorded mobile phone use. We used multiple linear regression models to assess possible predictors of the average number and duration of calls per day and logistic regression models to assess possible predictors of overestimation. The cumulative number and duration of calls as well as the time since first subscription of mobile phones were overestimated on average by the study participants. Likelihood to overestimate number and duration of calls was not significantly different for controls compared to cases (OR=1.1, 95%-CI: 0.5 to 2.5 and OR=1.9, 95%-CI: 0.85 to 4.3, respectively). However, likelihood to overestimate was associated with other health related factors such as age and sex. As a consequence, such factors act as confounders in studies relying solely on self-reported mobile phone use and have to be considered in the analysis.

Frequency of and predictors for patient-reported medical and medication errors in Switzerland

OBJECTIVES: To analyse the frequency of and identify risk factors for patient-reported medical errors in Switzerland. The joint effect of risk factors on error-reporting probability was modelled for hypothetical patients. METHODS: A representative population sample of Swiss citizens (n = 1306) was surveyed as part of the Commonwealth Fund’s 2010 lnternational Survey of the General Public’s Views of their Health Care System’s Performance in Eleven Countries. Data on personal background, utilisation of health care, coordination of care problems and reported errors were assessed. Logistic regression analysis was conducted to identify risk factors for patients’ reports of medical mistakes and medication errors. RESULTS: 11.4% of participants reported at least one error in their care in the previous two years (8% medical errors, 5.3% medication errors). Poor coordination of care experiences was frequent. 7.8% experienced that test results or medical records were not available, 17.2% received conflicting information from care providers and 11.5% reported that tests were ordered although they had been done before. Age (OR = 0.98, p = 0.014), poor health (OR = 2.95, p = 0.007), utilisation of emergency care (OR = 2.45, p = 0.003), inpatient-stay (OR = 2.31, p = 0.010) and poor care coordination (OR = 5.43, p <0.001) are important predictors for reporting error. For high utilisers of care that unify multiple risk factors the probability that errors are reported rises up to p = 0.8. CONCLUSIONS: Patient safety remains a major challenge for the Swiss health care system. Despite the health related and economic burden associated with it, the widespread experience of medical error in some subpopulations also has the potential to erode trust in the health care system as a whole.