Home About us Contact
|
Home About us Contact | ||
|
|
||
Predictive Accuracy (predictive + accuracy)
Selected AbstractsSurvival Model Predictive Accuracy and ROC CurvesBIOMETRICS, Issue 1 2005Patrick J. Heagerty Summary The predictive accuracy of a survival model can be summarized using extensions of the proportion of variation explained by the model, or R2, commonly used for continuous response models, or using extensions of sensitivity and specificity, which are commonly used for binary response models. In this article we propose new time-dependent accuracy summaries based on time-specific versions of sensitivity and specificity calculated over risk sets. We connect the accuracy summaries to a previously proposed global concordance measure, which is a variant of Kendall's tau. In addition, we show how standard Cox regression output can be used to obtain estimates of time-dependent sensitivity and specificity, and time-dependent receiver operating characteristic (ROC) curves. Semiparametric estimation methods appropriate for both proportional and nonproportional hazards data are introduced, evaluated in simulations, and illustrated using two familiar survival data sets. [source] Evaluation of molecular forms of prostate-specific antigen and human kallikrein 2 in predicting biochemical failure after radical prostatectomyINTERNATIONAL JOURNAL OF CANCER, Issue 3 2009Sven Wenske Abstract Most pretreatment risk-assessment models to predict biochemical recurrence (BCR) after radical prostatectomy (RP) for prostate cancer rely on total prostate-specific antigen (PSA), clinical stage, and biopsy Gleason grade. We investigated whether free PSA (fPSA) and human glandular kallikrein-2 (hK2) would enhance the predictive accuracy of this standard model. Preoperative serum samples and complete clinical data were available for 1,356 patients who underwent RP for localized prostate cancer from 1993 to 2005. A case-control design was used, and conditional logistic regression models were used to evaluate the association between preoperative predictors and BCR after RP. We constructed multivariable models with fPSA and hK2 as additional preoperative predictors to the base model. Predictive accuracy was assessed with the area under the ROC curve (AUC). There were 146 BCR cases; the median follow up for patients without BCR was 3.2 years. Overall, 436 controls were matched to 146 BCR cases. The AUC of the base model was 0.786 in the entire cohort; adding fPSA and hK2 to this model enhanced the AUC to 0.798 (p = 0.053), an effect largely driven by fPSA. In the subgroup of men with total PSA ,10 ng/ml (48% of cases), adding fPSA and hK2 enhanced the AUC of the base model to a similar degree (from 0.720 to 0.726, p = 0.2). fPSA is routinely measured during prostate cancer detection. We suggest that the role of fPSA in aiding preoperative prediction should be investigated in further cohorts. © 2008 Wiley-Liss, Inc. [source] Prognostic models in cirrhotics admitted to intensive care units better predict outcome when assessed at 48 h after admissionJOURNAL OF GASTROENTEROLOGY AND HEPATOLOGY, Issue 8pt1 2008Evangelos Cholongitas Abstract Background and Aim:, The accuracy of prognostic models in critically ill cirrhotics at admission to intensive care units (ICU) may be unreliable. Predictive accuracy could be improved by evaluating changes over time, but this has not been published. The aim of the present study was to assess the performance of prognostic models in cirrhotics at admission (baseline) and at 48 h to predict mortality in the ICU or within 6 weeks after discharge from the ICU. Methods:, One hundred and twenty-eight cirrhotics (77 males, mean age 49 ± 11.3 years) were consecutively admitted and alive 48 h after admission with 89% on mechanical ventilation, 76% on inotrope support, and 42% with renal failure. Prognostic models used were Child-Turcotte-Pugh (CTP), Model for End-stage Liver Disease (MELD), Acute Physiology and Chronic Health Evaluation (APACHE) II, Sequential Organ Failure Assessment (SOFA), failing organ systems (FOS) at baseline and at 48 h, ,score (difference between baseline and at 48 h) and the mean score (MN , score admission + 48 h/2) which were compared by area under the receiver operating characteristic curves (AUC). Results:, Mortality was 54.7% (n = 70) due to multiple organ failure in 55%. CTP, MELD, APACHE II, SOFA and FOS performed better at 48 h (AUC: 0.78, 0.86, 0.78, 0.88 and 0.85, respectively) than at baseline (AUC: 0.75, 0.78, 0.75, 0.81 and 0.79, respectively). The mean score had better discrimination than the baseline score; the ,score had poor predictive ability (AUC < 0.70). SOFA score (48 h: 0.88, mean: 0.88) and FOS (mean: 0.88) had the best accuracy, with a SOFA and MN-SOFA , 10 predicting mortality in 93% and 91%, respectively, and MN-FOS , 1.5 in 98%. Conclusions:, In cirrhotics, prognostic scores in the ICU at 48 h had better discrimination than baseline scores for short-term mortality. SOFA and FOS models had the best performance. [source] Emergency Thoracic Ultrasound in the Differentiation of the Etiology of Shortness of Breath (ETUDES): Sonographic B-lines and N-terminal Pro-brain-type Natriuretic Peptide in Diagnosing Congestive Heart FailureACADEMIC EMERGENCY MEDICINE, Issue 3 2009Andrew S. Liteplo MD Abstract Objectives:, Sonographic thoracic B-lines and N-terminal pro-brain-type natriuretic peptide (NT-ProBNP) have been shown to help differentiate between congestive heart failure (CHF) and chronic obstructive pulmonary disease (COPD). The authors hypothesized that ultrasound (US) could be used to predict CHF and that it would provide additional predictive information when combined with NT-ProBNP. They also sought to determine optimal two- and eight-zone scanning protocols when different thresholds for a positive scan were used. Methods:, This was a prospective, observational study of a convenience sample of adult patients presenting to the emergency department (ED) with shortness of breath. Each patient had an eight-zone thoracic US performed by one of five sonographers, and serum NT-ProBNP levels were measured. Chart review by two physicians blinded to the US results served as the criterion standard. The operating characteristics of two- and eight-zone thoracic US alone, compared to, and combined with NT-ProBNP test results for predicting CHF were calculated using both dichotomous and interval likelihood ratios (LRs). Results:, One-hundred patients were enrolled. Six were excluded because of incomplete data. Results of 94 patients were analyzed. A positive eight-zone US, defined as at least two positive zones on each side, had a positive likelihood ratio (LR+) of 3.88 (99% confidence interval [CI] = 1.55 to 9.73) and a negative likelihood ratio (LR,) of 0.5 (95% CI = 0.30 to 0.82), while the NT-ProBNP demonstrated a LR+ of 2.3 (95% CI = 1.41 to 3.76) and LR, of 0.24 (95% CI = 0.09 to 0.66). Using interval LRs for the eight-zone US test alone, the LR for a totally positive test (all eight zones positive) was infinite and for a totally negative test (no zones positive) was 0.22 (95% CI = 0.06 to 0.80). For two-zone US, interval LRs were 4.73 (95% CI = 2.10 to 10.63) when inferior lateral zones were positive bilaterally and 0.3 (95% CI = 0.13 to 0.71) when these were negative. These changed to 8.04 (95% CI = 1.76 to 37.33) and 0.11 (95% CI = 0.02 to 0.69), respectively, when congruent with NT-ProBNP. Conclusions:, Bedside thoracic US for B-lines can be a useful test for diagnosing CHF. Predictive accuracy is greatly improved when studies are totally positive or totally negative. A two-zone protocol performs similarly to an eight-zone protocol. Thoracic US can be used alone or can provide additional predictive power to NT-ProBNP in the immediate evaluation of dyspneic patients presenting to the ED. [source] High-Dimensional Cox Models: The Choice of Penalty as Part of the Model Building ProcessBIOMETRICAL JOURNAL, Issue 1 2010Axel Benner Abstract The Cox proportional hazards regression model is the most popular approach to model covariate information for survival times. In this context, the development of high-dimensional models where the number of covariates is much larger than the number of observations ( ) is an ongoing challenge. A practicable approach is to use ridge penalized Cox regression in such situations. Beside focussing on finding the best prediction rule, one is often interested in determining a subset of covariates that are the most important ones for prognosis. This could be a gene set in the biostatistical analysis of microarray data. Covariate selection can then, for example, be done by L1 -penalized Cox regression using the lasso (Tibshirani (1997). Statistics in Medicine16, 385,395). Several approaches beyond the lasso, that incorporate covariate selection, have been developed in recent years. This includes modifications of the lasso as well as nonconvex variants such as smoothly clipped absolute deviation (SCAD) (Fan and Li (2001). Journal of the American Statistical Association96, 1348,1360; Fan and Li (2002). The Annals of Statistics30, 74,99). The purpose of this article is to implement them practically into the model building process when analyzing high-dimensional data with the Cox proportional hazards model. To evaluate penalized regression models beyond the lasso, we included SCAD variants and the adaptive lasso (Zou (2006). Journal of the American Statistical Association101, 1418,1429). We compare them with "standard" applications such as ridge regression, the lasso, and the elastic net. Predictive accuracy, features of variable selection, and estimation bias will be studied to assess the practical use of these methods. We observed that the performance of SCAD and adaptive lasso is highly dependent on nontrivial preselection procedures. A practical solution to this problem does not yet exist. Since there is high risk of missing relevant covariates when using SCAD or adaptive lasso applied after an inappropriate initial selection step, we recommend to stay with lasso or the elastic net in actual data applications. But with respect to the promising results for truly sparse models, we see some advantage of SCAD and adaptive lasso, if better preselection procedures would be available. This requires further methodological research. [source] Improved prediction of recurrence after curative resection of colon carcinoma using tree-based risk stratificationCANCER, Issue 5 2004Martin Radespiel-Tröger M.D. Abstract BACKGROUND Patients who are at high risk of recurrence after undergoing curative (R0) resection for colon carcinoma may benefit most from adjuvant treatment and from intensive follow-up for early detection and treatment of recurrence. However, in light of new clinical evidence, there is a need for continuous improvement in the calculation of the risk of recurrence. METHODS Six hundred forty-one patients with R0-resected colon carcinoma who underwent surgery between January 1, 1984 and December 31, 1996 were recruited from the Erlangen Registry of Colorectal Carcinoma. The study end point was time until first locoregional or distant recurrence. The factors analyzed were: age, gender, site in colon, International Union Against Cancer (UICC) pathologic tumor classification (pT), UICC pathologic lymph node classification, histologic tumor type, malignancy grade, lymphatic invasion, venous invasion, number of examined lymph nodes, number of lymph node metastases, emergency presentation, intraoperative tumor cell spillage, surgeon, and time period. The resulting prognostic tree was evaluated by means of an independent sample using a measure of predictive accuracy based on the Brier score for censored data. Predictive accuracy was compared with several proposed stage groupings. RESULTS The prognostic tree contained the following variables: pT, the number of lymph node metastases, venous invasion, and emergency presentation. Predictive accuracy based on the validation sample was 0.230 (95% confidence interval [95% CI], 0.227,0.233) for the prognostic tree and 0.212 (95% CI, 0.209,0.215) for the UICC TNM sixth edition stage grouping. CONCLUSIONS The prognostic tree showed superior predictive accuracy when it was validated using an independent sample. It is interpreted easily and may be applied under clinical circumstances. Provided that their classification system can be validated successfully in other centers, the authors propose using the prognostic tree as a starting point for studies of adjuvant treatment and follow-up strategies. Cancer 2004;100:958,67. © 2004 American Cancer Society. [source] Validation and Clinical Utility of a Simple In-Home Testing Tool for Sleep-Disordered Breathing and Arrhythmias in Heart Failure: Results of the Sleep Events, Arrhythmias, and Respiratory Analysis in Congestive Heart Failure (SEARCH) StudyCONGESTIVE HEART FAILURE, Issue 5 2006William T. Abraham MD Fifty patients with New York Heart Association class III systolic heart failure were enrolled in this prospective multicenter study that compared the diagnostic accuracy of a home-based cardiorespiratory testing system with standard attended polysomnography. Patients underwent at least 2 nights of evaluation and were scored by blinded observers. At diagnostic cutoff points of ,5, ,10, and ,15 events per hour for respiratory disturbance severity, polysomnography demonstrated a sleep-disordered breathing prevalence of 69%, 59%, and 49%, respectively. Compared with polysomnography, the cardiorespiratory testing system demonstrated predictive accuracies of 73%, 73%, and 75%, which improved to 87%, 87%, and 83%, respectively, when analysis of covariance suggested reanalysis omitting one site's data. The system accurately identified both suspected and unsuspected arrhythmias. The device was judged by 80% of patients to be easy or very easy to use, and 74% of patients expressed a preference for the in-home system. Therefore, this system represents a reasonable home testing device in these patients. [source] N-Ace: Using solvent accessibility and physicochemical properties to identify protein N-acetylation sitesJOURNAL OF COMPUTATIONAL CHEMISTRY, Issue 15 2010Tzong-Yi Lee Abstract Protein acetylation, which is catalyzed by acetyltransferases, is a type of post-translational modification and crucial to numerous essential biological processes, including transcriptional regulation, apoptosis, and cytokine signaling. As the experimental identification of protein acetylation sites is time consuming and laboratory intensive, several computational approaches have been developed for identifying the candidates of experimental validation. In this work, solvent accessibility and the physicochemical properties of proteins are utilized to identify acetylated alanine, glycine, lysine, methionine, serine, and threonine. A two-stage support vector machine was applied to learn the computational models with combinations of amino acid sequences, and the accessible surface area and physicochemical properties of proteins. The predictive accuracy thus achieved is 5% to 14% higher than that of models trained using only amino acid sequences. Additionally, the substrate specificity of the acetylated site was investigated in detail with reference to the subcellular colocalization of acetyltransferases and acetylated proteins. The proposed method, N-Ace, is evaluated using independent test sets in various acetylated residues and predictive accuracies of 90% were achieved, indicating that the performance of N-Ace is comparable with that of other acetylation prediction methods. N-Ace not only provides a user-friendly input/output interface but also is a creative method for predicting protein acetylation sites. This novel analytical resource is now freely available at http://N-Ace.mbc.NCTU.edu.tw/. © 2010 Wiley Periodicals, Inc. J Comput Chem, 2010 [source] Evaluation Metrics in Classification: A Quantification of Distance-BiasCOMPUTATIONAL INTELLIGENCE, Issue 3 2003Ricardo Vilalta This article provides a characterization of bias for evaluation metrics in classification (e.g., Information Gain, Gini, ,2, etc.). Our characterization provides a uniform representation for all traditional evaluation metrics. Such representation leads naturally to a measure for the distance between the bias of two evaluation metrics. We give a practical value to our measure by observing the distance between the bias of two evaluation metrics and its correlation with differences in predictive accuracy when we compare two versions of the same learning algorithm that differ in the evaluation metric only. Experiments on real-world domains show how the expectations on accuracy differences generated by the distance-bias measure correlate with actual differences when the learning algorithm is simple (e.g., search for the best single feature or the best single rule). The correlation, however, weakens with more complex algorithms (e.g., learning decision trees). Our results show how interaction among learning components is a key factor to understand learning performance. [source] Employee Stock Option Fair-Value Estimates: Do Managerial Discretion and Incentives Explain Accuracy?,CONTEMPORARY ACCOUNTING RESEARCH, Issue 4 2006Leslie Hodder Abstract We examine the determinants of managers' use of discretion over employee stock option (ESO) valuation-model inputs that determine ESO fair values. We also explore the consequences of such discretion. Firms exercise considerable discretion over all model inputs, and this discretion results in material differences in ESO fair-value estimates. Contrary to conventional wisdom, we find that a large proportion of firms exercise value-increasing discretion. Importantly, we find that using discretion improves predictive accuracy for about half of our sample firms. Moreover, we find that both opportunistic and informational managerial incentives together explain the accuracy of firms' ESO fair-value estimates. Partitioning on the direction of discretion improves our understanding of managerial incentives. Our analysis confirms that financial statement readers can use mandated contextual disclosures to construct powerful ex ante predictions of ex post accuracy. [source] The validity of the Violence Risk Appraisal Guide (VRAG) in predicting criminal recidivismCRIMINAL BEHAVIOUR AND MENTAL HEALTH, Issue 2 2007Carolin Kröner Introduction,The VRAG is an actuarial risk assessment instrument, developed in Canada as an aid to estimating the probability of reoffending by mentally ill offenders. Aim,To test the predictive validity of the VRAG with a German sample. Method,The predictive validity of the VRAG was tested on a sample of 136 people charged with a criminal offence and under evaluation for criminal responsibility in the forensic psychiatry department at the University of Munich in 1994,95. The predicted outcome was tested by means of ROC analysis for correlation with the observed rate of recidivism between discharge after the 1994,95 assessment and the census date of 31 March 2003. Recidivism rate was calculated from the official records of the National Conviction Registry. Results,Just over 38% of the sample had reoffended by 2003. Their mean time-at-risk was 58 months (SD 3.391; range 0,115 months). The VRAG yielded a high predictive accuracy in the ROC analysis with an AUC of 0.703. For a constant time-at-risk < = 7 years, the predicted probability and observed rates of recidivism correlated significantly with Pearson's r = 0.941. Conclusions,The validity of the VRAG was replicated with a German sample. The VRAG yielded good predictive accuracy, despite differences in sample and outcome variables compared with its original sample. Copyright © 2007 John Wiley & Sons, Ltd. [source] The effect of discordance among violence and general recidivism risk estimates on predictive accuracyCRIMINAL BEHAVIOUR AND MENTAL HEALTH, Issue 3 2006Jeremy F. Mills Introduction,Previous research has shown that the prediction of short-term inpatient violence is negatively affected when clinicians' inter-rater agreement is low and when confidence in the estimate of risk is low. This study examined the effect of discordance between risk assessment instruments used to predict long-term general and violence risk in offenders. Methods,The Psychopathy Checklist , Revised (PCL,R), Level of Service Inventory , Revised (LSI,R), Violence Risk Appraisal Guide (VRAG), and the General Statistical Information on Recidivism (GSIR) were the four risk-prediction instruments used to predict post-release general and violent recidivism within a sample of 209 offenders. Results,The findings lend empirical support to the assumption that predictive accuracy is threatened where there is discordance between risk estimates. Discordance between instruments had the impact of reducing predictive accuracy for all instruments except the GSIR. Further, the influence of discordance was shown to be greater on certain instruments over others. Discordance had a moderating effect on both the PCL,R and LSI,R but not on the VRAG and GSIR. Conclusions,There is a distinct advantage when attempting to predict recidivism to employing measures such as the LSI-R, which includes dynamic variables and intervention-related criminogenic domains, over a measure purely of fixed characteristics, such as the GSIR; however, if there is discordance between the risk estimates, caution should be exercised and more reliance on the more static historically based instrument may be indicated. Copyright © 2006 John Wiley & Sons, Ltd. [source] Prediction of protein structural class by amino acid and polypeptide compositionFEBS JOURNAL, Issue 17 2002Rui-yan Luo A new approach of predicting structural classes of protein domain sequences is presented in this paper. Besides the amino acid composition, the composition of several dipeptides, tripeptides, tetrapeptides, pentapeptides and hexapeptides are taken into account based on the stepwise discriminant analysis. The result of jackknife test shows that this new approach can lead to higher predictive sensitivity and specificity for reduced sequence similarity datasets. Considering the dataset PDB40-B constructed by Brenner and colleagues, 75.2% protein domain sequences are correctly assigned in the jackknife test for the four structural classes: all-,, all-,, ,/, and , + ,, which is improved by 19.4% in jackknife test and 25.5% in resubstitution test, in contrast with the component-coupled algorithm using amino acid composition alone (AAC approach) for the same dataset. In the cross-validation test with dataset PDB40-J constructed by Park and colleagues, more than 80% predictive accuracy is obtained. Furthermore, for the dataset constructed by Chou and Maggiona, the accuracy of 100% and 99.7% can be easily achieved, respectively, in the resubstitution test and in the jackknife test merely taking the composition of dipeptides into account. Therefore, this new method provides an effective tool to extract valuable information from protein sequences, which can be used for the systematic analysis of small or medium size protein sequences. The computer programs used in this paper are available on request. [source] Modelling patterned ground distribution in Finnish Lapland: an integration of topographical, ground and remote sensing informationGEOGRAFISKA ANNALER SERIES A: PHYSICAL GEOGRAPHY, Issue 1 2006Jan Hjort Abstract New data technologies and modelling methods have gained more attention in the field of periglacial geomorphology during the last decade. In this paper we present a new modelling approach that integrates topographical, ground and remote sensing information in predictive geomorphological mapping using generalized additive modelling (GAM). First, we explored the roles of different environmental variable groups in determining the occurrence of non-sorted and sorted patterned ground in a fell region of 100 km2 at the resolution of 1 ha in northern Finland. Second, we compared the predictive accuracy of ground-topography- and remote-sensing-based models. The results indicate that non-sorted patterned ground is more common at lower altitudes where the ground moisture and vegetation abundance is relatively high, whereas sorted patterned ground is dominant at higher altitudes with relatively high slope angle and sparse vegetation cover. All modelling results were from good to excellent in model evaluation data using the area under the curve (AUC) values, derived from receiver operating characteristic (ROC) plots. Generally, models built with remotely sensed data were better than ground-topography-based models and combination of all environmental variables improved the predictive ability of the models. This paper confirms the potential utility of remote sensing information for modelling patterned ground distribution in subarctic landscapes. [source] BIOMOD , optimizing predictions of species distributions and projecting potential future shifts under global changeGLOBAL CHANGE BIOLOGY, Issue 10 2003Wilfried ThuillerArticle first published online: 9 OCT 200 Abstract A new computation framework (BIOMOD: BIOdiversity MODelling) is presented, which aims to maximize the predictive accuracy of current species distributions and the reliability of future potential distributions using different types of statistical modelling methods. BIOMOD capitalizes on the different techniques used in static modelling to provide spatial predictions. It computes, for each species and in the same package, the four most widely used modelling techniques in species predictions, namely Generalized Linear Models (GLM), Generalized Additive Models (GAM), Classification and Regression Tree analysis (CART) and Artificial Neural Networks (ANN). BIOMOD was applied to 61 species of trees in Europe using climatic quantities as explanatory variables of current distributions. On average, all the different modelling methods yielded very good agreement between observed and predicted distributions. However, the relative performance of different techniques was idiosyncratic across species, suggesting that the most accurate model varies between species. The results of this evaluation also highlight that slight differences between current predictions from different modelling techniques are exacerbated in future projections. Therefore, it is difficult to assess the reliability of alternative projections without validation techniques or expert opinion. It is concluded that rather than using a single modelling technique to predict the distribution of several species, it would be more reliable to use a framework assessing different models for each species and selecting the most accurate one using both evaluation methods and expert knowledge. [source] Upper digestive bleeding in cirrhosis.HEPATOLOGY, Issue 3 2003Post-therapeutic outcome, prognostic indicators Several treatments have been proven to be effective for variceal bleeding in patients with cirrhosis. The aim of this multicenter, prospective, cohort study was to assess how these treatments are used in clinical practice and what are the posttherapeutic prognosis and prognostic indicators of upper digestive bleeding in patients with cirrhosis. A training set of 291 and a test set of 174 bleeding cirrhotic patients were included. Treatment was according to the preferences of each center and the follow-up period was 6 weeks. Predictive rules for 5-day failure (uncontrolled bleeding, rebleeding, or death) and 6-week mortality were developed by the logistic model in the training set and validated in the test set. Initial treatment controlled bleeding in 90% of patients, including vasoactive drugs in 27%, endoscopic therapy in 10%, combined (endoscopic and vasoactive) in 45%, balloon tamponade alone in 1%, and none in 17%. The 5-day failure rate was 13%, 6-week rebleeding was 17%, and mortality was 20%. Corresponding findings for variceal versus nonvariceal bleeding were 15% versus 7% (P = .034), 19% versus 10% (P = .019), and 20% versus 15% (P = .22). Active bleeding on endoscopy, hematocrit levels, aminotransferase levels, Child-Pugh class, and portal vein thrombosis were significant predictors of 5-day failure; alcohol-induced etiology, bilirubin, albumin, encephalopathy, and hepatocarcinoma were predictors of 6-week mortality. Prognostic reassessment including blood transfusions improved the predictive accuracy. All the developed prognostic models were superior to the Child-Pugh score. In conclusion, prognosis of digestive bleeding in cirrhosis has much improved over the past 2 decades. Initial treatment stops bleeding in 90% of patients. Accurate predictive rules are provided for early recognition of high-risk patients. [source] Global potential distribution of an invasive species, the yellow crazy ant (Anoplolepis gracilipes) under climate changeINTEGRATIVE ZOOLOGY (ELECTRONIC), Issue 3 2008Youhua CHEN Abstract Changes to the Earth's climate may affect the distribution of countless species. Understanding the potential distribution of known invasive species under an altered climate is vital to predicting impacts and developing management policy. The present study employs ecological niche modeling to construct the global potential distribution range of the yellow crazy ant (Anoplolepis gracilipes) using past, current and future climate scenarios. Three modeling algorithms, GARP, BioClim and Environmental Distance, were used in a comparative analysis. Output from the models suggest firstly that this insect originated from south Asia, expanded into Europe and then into Afrotropical regions, after which it formed its current distribution. Second, the invasive risk of A. gracilipes under future climatic change scenarios will become greater because of an extension of suitable environmental conditions in higher latitudes. Third, when compared to the GARP model, BioClim and Environmental Distance models were better at modeling a species' ancestral distribution. These findings are discussed in light of the predictive accuracy of these models. [source] Evaluation of molecular forms of prostate-specific antigen and human kallikrein 2 in predicting biochemical failure after radical prostatectomyINTERNATIONAL JOURNAL OF CANCER, Issue 3 2009Sven Wenske Abstract Most pretreatment risk-assessment models to predict biochemical recurrence (BCR) after radical prostatectomy (RP) for prostate cancer rely on total prostate-specific antigen (PSA), clinical stage, and biopsy Gleason grade. We investigated whether free PSA (fPSA) and human glandular kallikrein-2 (hK2) would enhance the predictive accuracy of this standard model. Preoperative serum samples and complete clinical data were available for 1,356 patients who underwent RP for localized prostate cancer from 1993 to 2005. A case-control design was used, and conditional logistic regression models were used to evaluate the association between preoperative predictors and BCR after RP. We constructed multivariable models with fPSA and hK2 as additional preoperative predictors to the base model. Predictive accuracy was assessed with the area under the ROC curve (AUC). There were 146 BCR cases; the median follow up for patients without BCR was 3.2 years. Overall, 436 controls were matched to 146 BCR cases. The AUC of the base model was 0.786 in the entire cohort; adding fPSA and hK2 to this model enhanced the AUC to 0.798 (p = 0.053), an effect largely driven by fPSA. In the subgroup of men with total PSA ,10 ng/ml (48% of cases), adding fPSA and hK2 enhanced the AUC of the base model to a similar degree (from 0.720 to 0.726, p = 0.2). fPSA is routinely measured during prostate cancer detection. We suggest that the role of fPSA in aiding preoperative prediction should be investigated in further cohorts. © 2008 Wiley-Liss, Inc. [source] Elucidation of a protein signature discriminating six common types of adenocarcinomaINTERNATIONAL JOURNAL OF CANCER, Issue 4 2007Gregory C. Bloom Abstract Pathologists are commonly facing the problem of attempting to identify the site of origin of a metastatic cancer when no primary tumor has been identified, yet few markers have been identified to date. Multitumor classifiers based on microarray based RNA expression have recently been described. Here we describe the first approximation of a tumor classifier based entirely on protein expression quantified by two-dimensional gel electrophoresis (2DE). The 2DE was used to analyze the proteomic expression pattern of 77 similarly appearing (using histomorphology) adenocarcinomas encompassing 6 types or sites of origin: ovary, colon, kidney, breast, lung and stomach. Discriminating sets of proteins were identified and used to train an artificial neural network (ANN). A leave-one-out cross validation (LOOCV) method was used to test the ability of the constructed network to predict the single held out sample from each iteration with a maximum predictive accuracy of 87% and an average predictive accuracy of 82% over the range of proteins chosen for its construction. These findings demonstrate the use of proteomics to construct a highly accurate ANN-based classifier for the detection of an individual tumor type, as well as distinguishing between 6 common tumor types in an unknown primary diagnosis setting. © 2006 Wiley-Liss, Inc. [source] Prospective cohort study comparing sequential organ failure assessment and acute physiology, age, chronic health evaluation III scoring systems for hospital mortality prediction in critically ill cirrhotic patientsINTERNATIONAL JOURNAL OF CLINICAL PRACTICE, Issue 2 2006Y-C Chen Summary The aim of the study was to evaluate the usefulness of sequential organ failure assessment (SOFA) and acute physiology, age, chronic health evaluation III (APACHE III) scoring systems obtained on the first day of intensive care unit (ICU) admission in predicting hospital mortality in critically ill cirrhotic patients. The study enrolled 102 cirrhotic patients consecutively admitted to ICU during a 1-year period. Twenty-five demographic, clinical and laboratory variables were analysed as predicators of survival. Information considered necessary to calculate the Child,Pugh, SOFA and APACHE III scores on the first day of ICU admission was also gathered. Overall hospital mortality was 68.6%. Multiple logistic regression analysis revealed that mean arterial pressure, SOFA and APACHE III scores were significantly related to prognosis. Goodness-of-fit was good for the SOFA and APACHE III models. Both predictive models displayed a similar degree of the best Youden index (0.68) and overall correctness (84%) of prediction. The SOFA and APACHE III models displayed good areas under the receiver,operating characteristic curve (0.917 ± 0.028 and 0.912 ± 0.029, respectively). Finally, a strong and significant positive correlation exists between SOFA and APACHE III scores for individual patients (r2 = 0.628, p < 0.001). This investigation confirms the grave prognosis for cirrhotic patients admitted to ICU. Both SOFA and APACHE III scores are excellent tools to predict the hospital mortality in critically ill cirrhotic patients. The overall predictive accuracy of SOFA and APACHE III is superior to that of Child,Pugh system. The role of these scoring systems in describing the dynamic aspects of clinical courses and allocating ICU resources needs to be clarified. [source] A Comparative Study of the Use of Four Fall Risk Assessment Tools on Acute Medical WardsJOURNAL OF AMERICAN GERIATRICS SOCIETY, Issue 6 2005Michael Vassallo FRCP Objectives: To compare the effectiveness of four falls risk assessment tools (STRATIFY, Downton, Tullamore, and Tinetti) by using them simultaneously in the same environment. Design: Prospective, open, observational study. Setting: Two acute medical wards admitting predominantly older patients. Participants: One hundred thirty-five patients, 86 female, mean age±standard deviation 83.8±8.01 (range 56,100). Measurements: A single clinician prospectively completed the four falls risk assessment tools. The extent of completion and time to complete each tool was recorded. Patients were followed until discharge, noting the occurrence of falls. The sensitivity, specificity, negative predictive accuracy, positive predictive accuracy, and total predictive accuracy were calculated. Results: The number of patients that the STRATIFY correctly identified (n=90) was significantly higher than the Downton (n=46; P<.001), Tullamore (n=66; P=.005), or Tinetti (n=52; P<.001) tools, but the STRATIFY had the poorest sensitivity (68.2%). The STRATIFY was also the only tool that could be fully completed in all patients (n=135), compared with the Downton (n=130; P=.06), Tullamore (n=130; P=.06), and Tinetti (n=17; P<.001). The time required to complete the STRATIFY tool (average 3.85 minutes) was significantly less than for the Downton (6.34 minutes; P<.001), Tinetti (7.4 minutes; P<.001), and Tullamore (6.25 minutes; P<.001). The Kaplan-Meier test showed that the STRATIFY (log rank P=.001) and Tullamore tools (log rank P<.001) were effective at predicting falls over the first week of admission. The Downton (log rank P=.46) and Tinetti tools (log rank P=.41) did not demonstrate this characteristic. Conclusion: Significant differences were identified in the performance and complexity between the four risk assessment tools studied. The STRATIFY tool was the shortest and easiest to complete and had the highest predictive value but the lowest sensitivity. [source] A Follow-up of Deinstitutionalized Men with Intellectual Disabilities and Histories of Antisocial BehaviourJOURNAL OF APPLIED RESEARCH IN INTELLECTUAL DISABILITIES, Issue 4 2004Vernon L. Quinsey Background, There is frequently great concern about the dangerousness of deinstitutionalized men with intellectual disabilities who have been institutionalized because they are considered to be at high risk for the commission of serious antisocial acts or sexual offending. Unfortunately, there is little information on whether changes in the behaviour of these men can be used to adjust supervision so as to manage risk. Methods, An appraisal of men with intellectual disabilities and histories of serious antisocial behaviours who were residing in institutions about to be closed led to a 16 month follow-up of 58 of these clients who had been transferred to community settings. Results, A total of 67% exhibited antisocial behaviour of some kind and 47% exhibited ,hands-on' violent or sexual misbehaviours directed toward other clients or staff. The Violent Risk Appraisal Guide was the best predictor of new violent or sexual incidents and a variety of other pre-release predictors were related to the likelihood of antisocial incidents of any kind. Overall predictive accuracy was moderate. A field trial showed that monthly staff ratings of client characteristics were related to antisocial incidents. Conclusions, These preliminary data indicate that measures of dynamic risk involving staff ratings are worth developing and evaluating. [source] Do we need land-cover data to model species distributions in Europe?JOURNAL OF BIOGEOGRAPHY, Issue 3 2004Wilfried Thuiller Abstract Aim, To assess the influence of land cover and climate on species distributions across Europe. To quantify the importance of land cover to describe and predict species distributions after using climate as the main driver. Location, The study area is Europe. Methods, (1) A multivariate analysis was applied to describe land-cover distribution across Europe and assess if the land cover is determined by climate at large spatial scales. (2) To evaluate the importance of land cover to predict species distributions, we implemented a spatially explicit iterative procedure to predict species distributions of plants (2603 species), mammals (186 species), breeding birds (440 species), amphibian and reptiles (143 species). First, we ran bioclimatic models using stepwise generalized additive models using bioclimatic variables. Secondly, we carried out a regression of land cover (LC) variables against residuals from the bioclimatic models to select the most relevant LC variables. Finally, we produced mixed models including climatic variables and those LC variables selected as decreasing the residual of bioclimatic models. Then we compared the explanatory and predictive power of the pure bioclimatic against the mixed model. Results, (1) At the European coarse resolution, land cover is mainly driven by climate. Two bioclimatic axes representing a gradient of temperature and a gradient of precipitation explained most variation of land-cover distribution. (2) The inclusion of land cover improved significantly the explanatory power of bioclimatic models and the most relevant variables across groups were those not explained or poorly explained by climate. However, the predictive power of bioclimatic model was not improved by the inclusion of LC variables in the iterative model selection process. Main conclusion, Climate is the major driver of both species and land-cover distributions over Europe. Yet, LC variables that are not explained or weakly associated with climate (inland water, sea or arable land) are interesting to describe particular mammal, bird and tree distributions. However, the addition of LC variables to pure bioclimatic models does not improve their predictive accuracy. [source] The Seasoned-Equity Issues of UK Firms: Market Reaction and Issuance Method ChoiceJOURNAL OF BUSINESS FINANCE & ACCOUNTING, Issue 1-2 2006Edel Barnes Abstract: This study examines the seasoned equity issues of companies traded on the London Stock Exchange. Recent regulatory changes have allowed UK firms more discretion in choice of issue approach, and this has led many firms to issue through placing in preference to a rights issue. Having first documented the trend towards increasing use of placings, we go on to identify an interesting subset of placings that are less likely to be anticipated by the market, and find a significant positive market reaction to such placings, which contrasts with the significant negative reaction we find for issues by rights. We also examine the choice of seasoned equity issuance method, focusing on the choice between placings versus rights issues. We develop a model to explain the choice of equity issue method that achieves a high level of predictive accuracy. [source] Assessment of Markers for Identifying Patients at Risk for Life-Threatening Arrhythmic Events in Brugada SyndromeJOURNAL OF CARDIOVASCULAR ELECTROPHYSIOLOGY, Issue 1 2005YOUICHI AJIRO M.D. Introduction: Risk stratification for life-threatening arrhythmic events in Brugada syndrome is not yet established. The aim of the present study was to examine the usefulness of various markers in predicting life-threatening arrhythmic events in the Brugada syndrome. Methods and Results: Forty-six patients with Brugada-type ECGs were categorized into the symptomatic (n = 28) and asymptomatic (n = 18) groups. Statistical analyses were performed with respect to the usefulness of the following markers: SCN5A mutation, pharmacologic challenge, ventricular fibrillation (VF) inducibility by programmed electrical stimulation, and late potential (LP) by signal-averaged ECG (SAECG). Comparison between the two groups revealed a significant difference only in LP positivity (92.6% vs 47.1%, P = 0.0004). The symptomatic group had significantly lower RMS40, longer LAS40, and longer fQRSd compared with the asymptomatic group. A significant difference was noted, especially RMS40. The positive predictive value, negative predictive value, and predictive accuracy when setting a cutoff value of 15 ,V were 92.0%, 78.9%, and 86.4%, respectively. Furthermore, patients with an RMS40 value <15 ,V (n = 25) showed significantly higher rates of VF recurrence compared with patients with an RMS40 value , 15 ,V (n = 19, P = 0.047). Conclusion: Regarding risk stratification for identifying high-risk patients in Brugada syndrome, only LP by SAECG was shown to be useful, suggesting the importance of RMS40 in predicting the history of life-threatening arrhythmic events and the recurrence of VF. [source] Using support vector machines for prediction of protein structural classes based on discrete wavelet transformJOURNAL OF COMPUTATIONAL CHEMISTRY, Issue 8 2009Jian-Ding Qiu Abstract The prediction of secondary structure is a fundamental and important component in the analytical study of protein structure and functions. How to improve the predictive accuracy of protein structural classification by effectively incorporating the sequence-order effects is an important and challenging problem. In this study, a new method, in which the support vector machine combines with discrete wavelet transform, is developed to predict the protein structural classes. Its performance is assessed by cross-validation tests. The predicted results show that the proposed approach can remarkably improve the success rates, and might become a useful tool for predicting the other attributes of proteins as well. © 2008 Wiley Periodicals, Inc. J Comput Chem 2009 [source] Identifying native-like protein structures using physics-based potentialsJOURNAL OF COMPUTATIONAL CHEMISTRY, Issue 1 2002Brian N. Dominy Abstract As the field of structural genomics matures, new methods will be required that can accurately and rapidly distinguish reliable structure predictions from those that are more dubious. We present a method based on the CHARMM gas phase implicit hydrogen force field in conjunction with a generalized Born implicit solvation term that allows one to make such discrimination. We begin by analyzing pairs of threaded structures from the EMBL database, and find that it is possible to identify the misfolded structures with over 90% accuracy. Further, we find that misfolded states are generally favored by the solvation term due to the mispairing of favorable intramolecular ionic contacts. We also examine 29 sets of 29 misfolded globin sequences from Levitt's "Decoys ,R' Us" database generated using a sequence homology-based method. Again, we find that discrimination is possible with approximately 90% accuracy. Also, even in these less distorted structures, mispairing of ionic contacts results in a more favorable solvation energy for misfolded states. This is also found to be the case for collapsed, partially folded conformations of CspA and protein G taken from folding free energy calculations. We also find that the inclusion of the generalized Born solvation term, in postprocess energy evaluation, improves the correlation between structural similarity and energy in the globin database. This significantly improves the reliability of the hypothesis that more energetically favorable structures are also more similar to the native conformation. Additionally, we examine seven extensive collections of misfolded structures created by Park and Levitt using a four-state reduced model also contained in the "Decoys ,R' Us" database. Results from these large databases confirm those obtained in the EMBL and misfolded globin databases concerning predictive accuracy, the energetic advantage of misfolded proteins regarding the solvation component, and the improved correlation between energy and structural similarity due to implicit solvation. Z-scores computed for these databases are improved by including the generalized Born implicit solvation term, and are found to be comparable to trained and knowledge-based scoring functions. Finally, we briefly explore the dynamic behavior of a misfolded protein relative to properly folded conformations. We demonstrate that the misfolded conformation diverges quickly from its initial structure while the properly folded states remain stable. Proteins in this study are shown to be more stable than their misfolded counterparts and readily identified based on energetic as well as dynamic criteria. In summary, we demonstrate the utility of physics-based force fields in identifying native-like conformations in a variety of preconstructed structural databases. The details of this discrimination are shown to be dependent on the construction of the structural database. © 2002 Wiley Periodicals, Inc. J Comput Chem 23: 147,160, 2002 [source] Accurate prediction of proton chemical shifts.JOURNAL OF COMPUTATIONAL CHEMISTRY, Issue 16 2001Abstract Forty-five proton chemical shifts in 14 aromatic molecules have been calculated at several levels of theory: Hartree,Fock and density functional theory with several different basis sets, and also second-order Mřller,Plesset (MP2) theory. To obtain consistent experimental data, the NMR spectra were remeasured on a 500 MHz spectrometer in CDCl3 solution. A set of 10 molecules without strong electron correlation effects was selected as the parametrization set. The calculated chemical shifts (relative to benzene) of 29 different protons in this set correlate very well with the experiment, and even better after linear regression. For this set, all methods perform roughly equally. The best agreement without linear regression is given by the B3LYP/TZVP method (rms deviation 0.060 ppm), although the best linear fit of the calculated shifts to experimental values is obtained for B3LYP/6-311++G**, with an rms deviation of only 0.037 ppm. Somewhat larger deviations were obtained for the second test set of 4 more difficult molecules: nitrobenzene, azulene, salicylaldehyde, and o -nitroaniline, characterized by strong electron correlation or resonance-assisted intramolecular hydrogen bonding. The results show that it is possible, at a reasonable cost, to calculate relative proton shieldings in a similar chemical environment to high accuracy. Our ultimate goal is to use calculated proton shifts to obtain constraints for local conformations in proteins; this requires a predictive accuracy of 0.1,0.2 ppm. © 2001 John Wiley & Sons, Inc. J Comput Chem 22: 1887,1895, 2001 [source] Measuring diagnostic and predictive accuracy in disease management: an introduction to receiver operating characteristic (ROC) analysisJOURNAL OF EVALUATION IN CLINICAL PRACTICE, Issue 2 2006Ariel Linden DrPH MS Abstract Diagnostic or predictive accuracy concerns are common in all phases of a disease management (DM) programme, and ultimately play an influential role in the assessment of programme effectiveness. Areas, such as the identification of diseased patients, predictive modelling of future health status and costs and risk stratification, are just a few of the domains in which assessment of accuracy is beneficial, if not critical. The most commonly used analytical model for this purpose is the standard 2 × 2 table method in which sensitivity and specificity are calculated. However, there are several limitations to this approach, including the reliance on a single defined criterion or cut-off for determining a true-positive result, use of non-standardized measurement instruments and sensitivity to outcome prevalence. This paper introduces the receiver operator characteristic (ROC) analysis as a more appropriate and useful technique for assessing diagnostic and predictive accuracy in DM. Its advantages include; testing accuracy across the entire range of scores and thereby not requiring a predetermined cut-off point, easily examined visual and statistical comparisons across tests or scores, and independence from outcome prevalence. Therefore the implementation of ROC as an evaluation tool should be strongly considered in the various phases of a DM programme. [source] Forecasting international bandwidth capabilityJOURNAL OF FORECASTING, Issue 4 2005Gary Madden Abstract M-competition studies provide a set of stylized recommendations to enhance forecast reliability. However, no single method dominates across series, leading to consideration of the relationship between selected data characteristics and the reliability of alternative forecast methods. This study conducts an analysis of predictive accuracy in relation to Internet bandwidth loads. Extrapolation techniques that perform best in M-competitions perform relatively poorly in predicting Internet bandwidth loads. Such performance is attributed to Internet bandwidth data exhibiting considerably less structure than M-competition data. Copyright © 2005 John Wiley & Sons, Ltd. [source] | |||||||||||||||||||||||||||||||||||||||||||||||||