Model Development Data Research Articles

Development and validation of Prediction models for Risks of complications in Early-onset Pre-eclampsia (PREP): a prospective cohort study.

The prognosis of early-onset pre-eclampsia (before 34 weeks' gestation) is variable. Accurate prediction of complications is required to plan appropriate management in high-risk women. To develop and validate prediction models for outcomes in early-onset pre-eclampsia. Prospective cohort for model development, with validation in two external data sets. Model development: 53 obstetric units in the UK. Model transportability: PIERS (Pre-eclampsia Integrated Estimate of RiSk for mothers) and PETRA (Pre-Eclampsia TRial Amsterdam) studies. Pregnant women with early-onset pre-eclampsia. Nine hundred and forty-six women in the model development data set and 850 women (634 in PIERS, 216 in PETRA) in the transportability (external validation) data sets. The predictors were identified from systematic reviews of tests to predict complications in pre-eclampsia and were prioritised by Delphi survey. The primary outcome was the composite of adverse maternal outcomes established using Delphi surveys. The secondary outcome was the composite of fetal and neonatal complications. We developed two prediction models: a logistic regression model (PREP-L) to assess the overall risk of any maternal outcome until postnatal discharge and a survival analysis model (PREP-S) to obtain individual risk estimates at daily intervals from diagnosis until 34 weeks. Shrinkage was used to adjust for overoptimism of predictor effects. For internal validation (of the full models in the development data) and external validation (of the reduced models in the transportability data), we computed the ability of the models to discriminate between those with and without poor outcomes (c-statistic), and the agreement between predicted and observed risk (calibration slope). The PREP-L model included maternal age, gestational age at diagnosis, medical history, systolic blood pressure, urine protein-to-creatinine ratio, platelet count, serum urea concentration, oxygen saturation, baseline treatment with antihypertensive drugs and administration of magnesium sulphate. The PREP-S model additionally included exaggerated tendon reflexes and serum alanine aminotransaminase and creatinine concentration. Both models showed good discrimination for maternal complications, with anoptimism-adjusted c-statistic of 0.82 [95% confidence interval (CI) 0.80 to 0.84] for PREP-L and 0.75 (95% CI 0.73 to 0.78) for the PREP-S model in the internal validation. External validation of the reduced PREP-L model showed good performance with a c-statistic of 0.81 (95% CI 0.77 to 0.85) in PIERS and 0.75 (95% CI 0.64 to 0.86) in PETRA cohorts for maternal complications, and calibrated well with slopes of 0.93 (95% CI 0.72 to 1.10) and 0.90 (95% CI 0.48 to 1.32), respectively. In the PIERS data set, the reduced PREP-S model had a c-statistic of 0.71 (95% CI 0.67 to 0.75) and a calibration slope of 0.67 (95% CI 0.56 to 0.79). Low gestational age at diagnosis, high urine protein-to-creatinine ratio, increased serum urea concentration, treatment with antihypertensive drugs, magnesium sulphate, abnormal uterine artery Doppler scan findings and estimated fetal weight below the 10th centile were associated with fetal complications. The PREP-L model provided individualised risk estimates in early-onset pre-eclampsia to plan management of high- or low-risk individuals. The PREP-S model has the potential to be used as a triage tool for risk assessment. The impacts of the model use on outcomes need further evaluation. Current Controlled Trials ISRCTN40384046. The National Institute for Health Research Health Technology Assessment programme.

Model Development and Experimental Data Analysis for Calcium Carbonate Membrane Scaling during Dead-End Filtration with Agitation

The development of reliable criteria for determining the onset of scaling of sparingly soluble salts on reverse osmosis desalination membranes is of great practical significance. Calcium carbonate, which is dealt with in this study, is a commonly encountered membrane scale compound. Reliable modeling and simulation tools of the reverse osmosis operation (accounting for onset and development of scaling) are very important for process optimization, aiming to reduce the effects of scaling and to minimize the use of antiscalants; such comprehensive modeling requires knowledge of key scaling parameters. In recent years a procedure has been developed to extract these parameters by employing a facile testing procedure, involving a small pressurized dead-end filtration cell. Using such a stirred test cell, the conditions are much closer to the actual reverse osmosis desalination process, compared to those in a non-agitated cell. Here, a detailed model of the testing process is developed and a procedure for estima...

Development of a Gestational Age–Specific Case Definition for Neonatal Necrotizing Enterocolitis

Necrotizing enterocolitis (NEC) is a major cause of neonatal morbidity and mortality. Preventive and therapeutic research, surveillance, and quality improvement initiatives are hindered by variations in case definitions. To develop a gestational age (GA)-specific case definition for NEC. We conducted a prospective 34-month population study using clinician-recorded findings from the UK National Neonatal Research Database between December 2011 and September 2014 across all 163 neonatal units in England. We split study data into model development and validation data sets and categorized GA into groups (group 1, less than 26 weeks' GA; group 2, 26 to less than 30 weeks' GA; group 3, 30 to less than 37 weeks' GA; group 4, 37 or more weeks' GA). We entered GA, birth weight z score, and clinical and abdominal radiography findings as candidate variables in a logistic regression model, performed model fitting 1000 times, averaged the predictions, and used estimates from the fitted model to develop an ordinal NEC score and cut points to develop a dichotomous case definition based on the highest area under the receiver operating characteristic curves [AUCs] and positive predictive values [PPVs]. Abdominal radiography performed to investigate clinical concerns. Ordinal NEC likelihood score, dichotomous case definition, and GA-specific probability plots. Of the 3866 infants, the mean (SD) birth weight was 2049.1 (1941.7) g and mean (SD) GA was 32 (5) weeks; 2032 of 3663 (55.5%) were male. The total included 2978 infants (77.0%) without NEC and 888 (23.0%) with NEC. Infants with NEC in group 1 were less likely to present with pneumatosis (31.1% vs 47.2%; P = .01), blood in stool (11.8% vs 29.6%; P < .001), or mucus in stool (2.1% vs 5.6%; P = .048) but more likely to present with gasless abdominal radiography findings (6.3% vs 0.9%; P = .009) compared with infants with NEC in group 3. In the ordinal NEC score analysis, we allocated 3 points to pneumatosis, 2 points to blood in stool, and 1 point each to abdominal tenderness and abdominal discoloration; 1 point was assigned if 1 or more of pneumoperitoneum, fixed loop, and portal venous gas were present, and 1 point was assigned if both increased and/or bilious aspirates and abdominal distension were present. The cutoff scores for the dichotomous GA-specific case definition were 2 or greater for infants in groups 1 and 2, 3 or greater for infants in group 3, and 4 or greater for infants in group 4. The ordinal NEC score and dichotomous case definition discriminated well between infants with (AUC, 87%) and without (AUC, 80%) NEC. The case definition has a sensitivity of 66.2% (95% CI, 63.0-69.4), a specificity of 94.4% (95% CI, 93.2-95.4), an AUC of 80.0% (95% CI, 79-82), and a PPV of 85.5% (95% CI, 82.6-88.1). Applying the cut points to the 431 infants who underwent a laparotomy yielded a sensitivity of 76.5% (95% CI, 70.0-82.1), a specificity of 74.4% (95% CI, 68.3-80.0), an AUC of 75.0% (95% CI, 71.0- 80.0), and a PPV of 72.9% (95% CI, 66.4-78.7). The risk of NEC and clinical presentation are associated with GA. Adoption of a consistent GA-specific case definition would strengthen global efforts to reduce the population burden of this devastating neonatal disease.

An Optimal Sample Data Usage Strategy to Minimize Overfitting and Underfitting Effects in Regression Tree Models Based on Remotely-Sensed Data

Regression tree models have been widely used for remote sensing-based ecosystem mapping. Improper use of the sample data (model training and testing data) may cause overfitting and underfitting effects in the model. The goal of this study is to develop an optimal sampling data usage strategy for any dataset and identify an appropriate number of rules in the regression tree model that will improve its accuracy and robustness. Landsat 8 data and Moderate-Resolution Imaging Spectroradiometer-scaled Normalized Difference Vegetation Index (NDVI) were used to develop regression tree models. A Python procedure was designed to generate random replications of model parameter options across a range of model development data sizes and rule number constraints. The mean absolute difference (MAD) between the predicted and actual NDVI (scaled NDVI, value from 0–200) and its variability across the different randomized replications were calculated to assess the accuracy and stability of the models. In our case study, a six-rule regression tree model developed from 80% of the sample data had the lowest MAD (MADtraining = 2.5 and MADtesting = 2.4), which was suggested as the optimal model. This study demonstrates how the training data and rule number selections impact model accuracy and provides important guidance for future remote-sensing-based ecosystem modeling.

Hybrid simulation modeling for humanitarian relief chain coordination

Purpose – The purpose of this paper is to present a conceptual framework for a hybrid simulation model that can be used to study the decision making and behaviors of humanitarian logistics actors to determine how/whether certain coordination mechanisms enable better relief chain efficiency and effectiveness over time. Design/methodology/approach – The agent-based portion of the model is used to represent human decision making and interactions in a more realistic way than has been done previously, and the discrete-event simulation (DES) portion of the model allows the movement of vehicles, materials, and information throughout a supply network to be represented in a way that allows for dynamic and stochastic behavior. Findings – Coordinated efforts by actors in humanitarian logistics operations involve complex interactions and adaptations over time, which can be capture and explored via hybrid agent-based model (ABM)-DES modeling. Research limitations/implications – This paper describes a framework for a hybrid ABM-DES model. The actual development and implementation of the model, including input data collection and analysis, model development, experimentation, and output data collection and analysis, will be the subject of future work. Practical implications – The hybrid model framework provides other researchers with a starting point for model development. Social implications – This paper provides a basis for future modeling and assessment of coordination in humanitarian logistics, an area that is in need of research. Originality/value – The hybrid simulation modeling framework presented in this paper is a novel application of a new modeling methodology to the problem of coordination in humanitarian logistics.

A Simple Nearest-Neighbor Technique to Predict the Soil Water Retention Curve

Abstract. Knowing the soil water retention curve (WRC) is essential for analyzing soil hydraulic behavior within the vadose zone. The van Genuchten (VG) soil hydraulic equation is one of the most frequently adopted models to parameterize the WRC. Some measured water retention points are needed to fit the VG model, but direct measurement of water content versus matric potential is expensive and time consuming. A pedotransfer function (PTF) enables indirect determination of a WRC from basic soil information. The typical method employed to derive a PTF using the VG model (VG-PTF) is to establish a mathematical relationship between the parameters of the VG model and basic soil data. However, both establishing and reusing a VG-PTF for new soils are challenging due to several reasons, such as over-parameterization, low correlation between basic soil data and VG parameters, and interdependency among parameters. In this study, a nonparametric approach based on the k nearest neighbor technique was designed and tested to establish a VG-PTF. A subset of soils from the UNSODA database (n = 554)) and a data set from Belgium (n = 69) were used as the model development and validation data sets, respectively. The proposed PTF showed reasonable accuracy and reliability and was comparable to well-known parametric VG-PTFs available in the literature.

Machine Learning for Predictive Modelling based on Small Data in Biomedical Engineering

Experimental datasets in bioengineering are commonly limited in size, thus rendering Machine Learning (ML) impractical for predictive modelling. Novel techniques of multiple runs for model development and surrogate data analysis for model validation are suggested for prediction of biomedical outcomes based on small datasets for classification and regression tasks. The proposed framework was applied to designing a Neural Network model for osteoarthritic bone fracture risk stratification, and a Decision Tree model for prediction of antibody-mediated kidney transplant rejection. Despite the small datasets (35 bone specimens and 80 kidney transplants), the two models achieved high accuracy of 98.3% and 85%, respectively.

Capacity and power fade cycle-life model for plug-in hybrid electric vehicle lithium-ion battery cells containing blended spinel and layered-oxide positive electrodes

This paper proposes and validates a semi-empirical cycle-life model for lithium-ion pouch cells containing blended spinel and layered-oxide positive electrodes. For the model development and validation experimental data obtained during an aging campaign is used. During the campaign the influence of charge sustaining/depleting operation, minimum state of charge (SOC), charging rate and temperature on the aging process is studied. The aging profiles, which are prescribed in power mode, are selected to be representative of realistic plug-in hybrid electric vehicle (PHEV) operation. The proposed model describes capacity fade and resistance increase as function of the influencing stress factors and battery charge throughput. Due to its simplicity but still good accuracy, the applications of the proposed aging model include the design of algorithms for battery state-of-health (SOH) monitoring and prognosis, PHEV optimal energy management including battery aging, and the study of aging propagation among battery cells in advanced energy storage systems.

비선형 회귀분석, 인공신경망, 구조방정식을 이용한 지방부 4지 신호교차로 교통사고 예측모형 성능 비교 연구

도로의 안전성을 평가하기 위한 방법으로서 교통사고 자료를 이용하는 방법, 사전-사후평가를 통한 방법 또는 전문가 의견이나 기존 문헌을 통한 방법 등 다양한 방법들이 존재한다. 특히, 교차로 교통 안전성을 평가하는 경우 많은 연구들이 교통사고예측모형 개발을 통하여 교통사고와 관련한 원인과 안전성을 평가하고 있다. 교통사고예측모형 개발에 있어서 모형의 예측력과 전용성을 확보하는 것이 중요하다. 즉, 예측력을 확보함으로써 교통사고 건수나 교통 안전성 판단의 지표를 예측하는데 오차를 줄일 수 있고, 전용성을 확보함으로써 개발된 모형이 다른 지점이나 구간에 적용하더라도 문제없이 적용될 수 있는 대표성을 가질 수 있다. 따라서 본 연구에서는 교통사고예측모형 개발에 주로 사용되는 회귀모형과 인공신경망, 구조방정식을 이용하여 교통사고예측모형을 각각 개발하였으며, 개발될 모형의 예측력과 전용성을 평균절대오차와 평균제곱예측오차를 기준으로 확인하였다. 90개소 신호교차로의 모형개발자료를 이용하여 세 가지 방법으로 교통사고예측모형을 개발 후 개발데이터를 통해 예측력을 비교한 결과 인공신경망이 가장 높은 예측력을 보였다. 또한 모형의 전용성 검증을 위하여 별도로 수집한 33개소 신호교차로의 모형검증자료를 이용하여 개발된 모형을 검증한 결과 비선형 회귀모형이 가장 적합한 것으로 나타났다. 모형개발 과정에서 가장 높은 예측력을 보인 인공신경망의 경우 다른 대상지에서 수집된 모형검증 자료를 적용하였을 때 예측력에 큰 변화를 보여 전용성이 떨어진 것으로 분석되었다. For the evaluation of roadway safety, diverse methods, including before-after studies, simple comparison using historic traffic accident data, methods based on experts' opinion or literature, have been applied. Especially, many research efforts have developed traffic accident prediction models in order to identify critical elements causing accidents and evaluate the level of safety. A traffic accident prediction model must secure predictability and transferability. By acquiring the predictability, the model can increase the accuracy in predicting the frequency of accidents qualitatively and quantitatively. By guaranteeing the transferability, the model can be used for other locations with acceptable accuracy. To this end, traffic accident prediction models using non-linear regression, artificial neural network, and structural equation were developed in this study. The predictability and transferability of three models were compared using a model development data set collected from 90 signalized intersections and a model validation data set from other 33 signalized intersections based on mean absolute deviation and mean squared prediction error. As a result of the comparison using the model development data set, the artificial neural network showed the highest predictability. However, the non-linear regression model was found out to be most appropriate in the comparison using the model validation data set. Conclusively, the artificial neural network has a strong ability in representing the relationship between the frequency of traffic accidents and traffic and road design elements. However, the predictability of the artificial neural network significantly decreased when the artificial neural network was applied to a new data which was not used in the model developing.

Outcome prediction model for severe traumatic brain injury

Treatment of severe traumatic brain injury is aided by better prediction of outcomes. The purpose of the present study was to develop and validate a prediction model using retrospective analysis of prospectively collected clinical data from two tertiary critical care medical centers in Japan. Data were collected from 253 patients with a Glasgow Coma Scale score of <9. Within 24 h of their admission, 15 factors possibly related to outcome were evaluated. The dataset was randomly split into training and validation datasets using the repeated random subsampling method. A logistic regression model was fitted to the training dataset and predictive accuracy was assessed using the validation data. The best model included the variables age, pupillary light reflex, extensive subarachnoid hemorrhage, intracranial pressure, and midline shift. The estimated area under the curve for the model development data was 0.957, with a 95% confidence interval of 0.926-0.987, and that for validation data was 0.947, with a 95% confidence interval of 0.909-0.980. Our predictive model was shown to have high predictive value. It will be useful for review of treatment, family counseling, and efficient allocation of resources for patients with severe traumatic brain injury.

Applicability of day-to-day variation in behavior for the automated detection of lameness in dairy cows

Lameness is a major problem in modern dairy husbandry and has welfare implications and other negative consequences. The behavior of dairy cows is influenced by lameness. Automated lameness detection can, among other methods, be based on day-to-day variation in animal behavior. Activity sensors that measure lying time, number of lying bouts, and other parameters were used to record behavior per cow per day. The objective of this research was to develop and validate a lameness detection model based on daily activity data. Besides the activity data, milking data and data from the computerized concentrate feeders were available as input data. Locomotion scores were available as reference data. Data from up to 100 cows collected at an experimental farm during 23mo in 2010 and 2011 were available for model development. Behavior is cow-dependent, and therefore quadratic trend models were fitted with a dynamic linear model on-line per cow for 7 activity variables and 2 other variables (milk yield per day and concentrate leftovers per day). It is assumed that lameness develops gradually; therefore, a lameness alert was given when the linear trend in 2 or more of the 9 models differed significantly from zero in a direction that corresponded with lameness symptoms. The developed model was validated during the first 4mo of 2012 with almost 100 cows on the same farm by generating lameness alerts each week. Performance on the model validation data set was comparable with performance on the model development data set. The overall sensitivity (percentage of detected lameness cases) was 85.5% combined with specificity (percentage of nonlame cow-days that were not alerted) of 88.8%. All variables contributed to this performance. These results indicate that automated lameness detection based on day-to-day variation in behavior is a useful tool for dairy management.

Development and Validation of Models to Predict Hospital Admission for Emergency Department Patients

Background: Boarding, or patients waiting to be admitted to hospital, has been shown as a significant contributing factor at overcrowding in emergency departments (ED). Predicting hospital admission at triage has been proposed as having the potential to help alleviate ED overcrowding. The objective of this paper is to develop and validate a model to predict hospital admission at triage to help alleviate ED overcrowding. Methods: Administrative records between April 1, 2010 and November 31, 2010 in an adult ED were used to derive and validate two prediction models, one based on Coxian phase type distribution (the PH model), the other based on logistic regression. Separate data sets were used for model development (data between April 1, 2010 and July 31, 2010) and validation (data between August 1, 2010 and November 31, 2010). Results: There were a total of 14,542 ED visits and 2,602 (17.89%) hospital admissions in the derivation cohort. In both models, acuity levels, model of arrival, and main reason of the visit are strong predictors of hospital admission; number of patients at the ED, as well as gender, are also predictors, albeit with ORs closer to 1. Patient age and timing of visits are not strong predictors. The PH model has an AUC of 0.89 compared with AUC of 0.83 for logistic regression model; with a cut- off value of 0.50, the PH model correctly predicted 86.3% of visits, compared to 84.4% for the logistic regression model. Results of the validation cohort were similar: the PH model has an AUC of 0.88, compared to AUC of 0.83 for the logistic model. Conclusions: PH and logistic models can be used to provide reasonably accurate prediction of hospital admission for ED patients, with the PH model offering more accurate predictions

Abstract P4-12-03: Towards a risk prediction model for breast cancer that utilizes breast tissue risk features

Abstract Background: Optimal early detection and prevention strategies for breast cancer are predicated on our ability to accurately identify women at increased risk for disease. Current breast cancer risk prediction models rely on clinical and epidemiologic features. As demonstrated with other cancers, such as those of the cervix and colon, cancer risk assessment is enhanced when the tissue at risk is examined for evidence of premalignant abnormalities. Here, we present a multivariate model for individualized breast cancer risk prediction that incorporates tissue features identified from biopsies from women with benign breast disease (BBD). Methods: We used our cohort of more than 9000 women diagnosed with BBD at the Mayo Clinic from 1967–1991, and identified a nested case:control series of 377 women who developed cancer (cases) matched with 734 women who did not (controls). We identified a set of variables that either alone or in two-way interactions contributed to a multivariable risk prediction model. We combined this relative risk model with the baseline incidence of breast cancer and the baseline hazard of death from our full BBD cohort to develop a risk prediction tool that computes absolute risk estimates for breast cancer. We calculated 5-year risk predictions for these cases and controls using our model and using the Gail model and compared c-statistics from these risk predictions. We validated our model using an independent set of 378 breast cancer cases and 728 matched controls drawn from our BBD cohort. Results: Cases and controls in the model development data set differed significantly on a number of features when examined individually. Those included in a multivariable relative risk model for breast cancer among women with BBD are shown below. After including interactions between lobular involution, and histologic impression, with age at first live birth/number of children, and degree of involution with sclerosing adenosis, we devised a model that predicts the absolute risk of breast cancer. Predicted five-year risks in the model development series, we observed a c-statistic of 0.66, higher than the c-statistic of 0.61 for the Gail model. With risk prediction estimates from our validation series of cases and matched controls, the c-statistic of our model decreased to 0.59 and the c-statistic for the Gail model was 0.55. Conclusion: We have identified an initial risk prediction model for breast cancer in women with BBD. The identified model includes basic demographic information as well as a number of histologic features of the tissue samples. Its initial performance is promising, providing improved discrimination over the Gail model in our replication data set. Further work is needed to identify features which might provide additional predictive information for breast cancer risk. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P4-12-03.

Detection of biomass change in a Norwegian mountain forest area using small footprint airborne laser scanner data

Different approaches for estimation of change in biomass between two points in time by means of airborne laser scanner data were tested. Both field and laser data were collected at two occasions on 52 sample plots in a mountain forest in southeastern Norway. In the first approach, biomass change was estimated as the difference between predicted biomass for the two measurement occasions. Joint models for the biomass at both occasions were fitted using different height and density variables from laser data as explanatory variables. The second approach modelled the observed change directly using the change in different variables extracted from the laser data as explanatory variables. In the third approach we modelled the relative change in biomass. The explanatory variables were also expressed as relative change between measurement occasions. In all approaches we allowed spline terms to be entered. We also investigated the aptness of models for which the residual variance was modeled by allowing it to be proportional to the area of the plot on which biomass was assessed. All alternative models were initially assessed by AIC. All models were also evaluated by estimating biomass change on the model development data. This evaluation indicated that the two direct approaches (approach 2 and 3) were better than relying on modeling biomass at both occasions and taking change as the difference between biomass estimates. Approach 2 seemed to be slightly better than approach 3 based on assessments of bias in the evaluation.

QSAR and QSPR Model Interpretation Using Partial Least Squares (PLS) Analysis

Carefully developed quantitative structure-activity and structure-property relationship models contain detailed information regarding how differences in the molecular structure of compounds correlate with differences in the observed biological or other physicochemical properties of those compounds. The ability to understand the behavior of existing molecules and to design new molecules is facilitated by using an objective method to extract and explain the details of the underlying structure-activity or structure-property relationship. Furthermore, a clear understanding of how and why compounds behave as they do can lead to new innovations through model-directed selection of compounds to be used in complex mixtures such as laundry detergents, fabric softeners, and shampoos. Such a method has been developed based on partial least-squares (PLS) regression analysis that allows for the identification of specific structural trends that relate to differences in observed properties. But the analysis of the completed model is only the last step of the process. The model development process itself affects the ability to extract a clear interpretation of the model. Everything from the selection of initial pool of molecular descriptors to evaluate to data set and model optimization impacts the ability to derive detailed molecular design information. This review describes the method details and examples of the use of PLS for model interpretation and also outlines suggestions regarding model development and model and data set optimization that enable the interpretation process.

Simulation-optimisation in modelling ionic polymer-metal composites actuators

The increasing pressure on the development time of new materials and devices has changed the modelling and design process over the years. In the past, they mainly consisted of experimentation and physical prototyping. Clearly, it is hard to incorporate changes in finished prototypes, while producing a variety of different prototypes at once may be very expensive. To this aim, computer simulation models such as circuit design models and continuous system simulation models are widely used in engineering modelling, design and analysis. The studies towards a better understanding of complex systems require quantitative model development, making optimisation and experimental data fitting tools indispensable. In this framework, the modelling of ionic polymer-metal composites (IPMCs) is studied. In particular, this paper deals with simulation-optimisation issues arising in the model calibration of a particular IPMC-based actuator in air. We consider a non-linear dynamical model of the device, with lumped parameters, able to estimate the IPMC actuator absorbed current, together with the mechanical quantities of interest, which, in the case under study, are the free deflection and/or the blocked force. Two optimisation problems have been formulated, focussing on different stages of the model parameters identification. The strategies adopted to solve the problems allow to achieve some promising – although preliminary – results.

Discharge destination following lower limb fracture: Development of a prediction model to assist with decision making

BackgroundAccurate prediction of the likelihood of discharge to inpatient rehabilitation following lower limb fracture made on admission to hospital may assist patient discharge planning and decrease the burden on the hospital system caused by delays in decision making. AimsTo develop a prognostic model for discharge to inpatient rehabilitation. MethodIsolated lower extremity fracture cases (excluding fractured neck of femur), captured by the Victorian Orthopaedic Trauma Outcomes Registry (VOTOR), were extracted for analysis. A training data set was created for model development and validation data set for evaluation. A multivariable logistic regression model was developed based on patient and injury characteristics. Models were assessed using measures of discrimination (C-statistic) and calibration (Hosmer–Lemeshow (H–L) statistic). ResultsA total of 1429 patients met the inclusion criteria and were randomly split into training and test data sets. Increasing age, more proximal fracture type, compensation or private fund source for the admission, metropolitan location of residence, not working prior to injury and having a self-reported pre-injury disability were included in the final prediction model. The C-statistic for the model was 0.92 (95% confidence interval (CI) 0.88, 0.95) with an H–L statistic of χ2=11.62, p=0.17. For the test data set, the C-statistic was 0.86 (95% CI 0.83, 0.90) with an H–L statistic of χ2=37.98, p<0.001. ConclusionA model to predict discharge to inpatient rehabilitation following lower limb fracture was developed with excellent discrimination although the calibration was reduced in the test data set. This model requires prospective testing but could form an integral part of decision making in regards to discharge disposition to facilitate timely and accurate referral to rehabilitation and optimise resource allocation.

Developing community infrastructure for hydrologic modeling

Third Workshop on the Community Hydrologic Modeling Platform (CHyMP): A Strategic and Implementation Plan; Irvine, California, 15–17 March 2011; The objective of the Community Hydrologic Modeling Platform (CHyMP), a grassroots effort begun in 2008, is to build the cyberinfrastructure and human infrastructure for community‐driven, integrated model development and comprehensive data set compilation, as well as a framework for model distribution, high‐performance computing access, and technical support. CHyMP recognizes that it is difficult to evaluate and access many existing hydrologic models and the data needed to calibrate them, and these shortcomings impede scientific progress. The third CHyMP workshop, “A Strategic and Implementation Plan,” funded by the U.S. National Science Foundation and the University of California Center for Hydrologic Modeling, brought together more than 30 participants from universities, government agencies, and the private sector to focus on defining steps to begin implementing CHyMP. Participants discussed four aspects critical to community modeling: the data required, the need for benchmarking and intercomparison exercises, the importance of evaluating platforms for such exercises, and the vision of a national water modeling framework.

Rhizoctonia Web Blight Development on Container-Grown Azalea in Relation to Time and Environmental Factors.

Rhizoctonia web blight, caused by binucleate Rhizoctonia spp., is an annual problem in the southern United States on container-grown azaleas (Rhododendron spp.) that receive daily irrigation. Disease progress was assessed weekly from mid-May to mid-September on nursery-grown plants at three locations in Mississippi and Alabama in 2006, 2007, and 2008. Disease onset, defined as the appearance of blighted leaves at the exterior canopy of at least one plant, occurred on average on 20 July, and calendar date was a more precise predictor of disease onset than several combined time-weather variables. Disease progress curves exhibited weekly fluctuations around a typically exponential increase in the mean number of symptomatic leaves per plant until early to mid-September, after which web blight severity leveled off or declined due to disease-induced leaf dehiscence and the appearance of new, asymptomatic leaves. Based on the relative increase in the log-transformed number of infected leaves per plant, weekly assessment periods were classified as having slow (≤0%), intermediate (>0 to <10%), or rapid (≥10% increase) disease progress. Three-day moving averages (MA) of various weather variables were calculated, and lagged values (by 5 days) of the MA were used in an attempt to predict disease progress as slow, intermediate, or rapid. Of the periods assessed as having slow disease progress in the 2006-2007 data set (model development data), 90.6% (29 of 32) met at least one of the following heuristically derived criteria for the lagged MA: min. temperature < 20.0°C, max. temperature > 35.0°C, avg. vapor pressure deficit < 2.50 hPa, or day of the year > 240 (28 August). One or more of these same criteria were met in 5 of 16 (31.2%) assessment periods with rapid disease progress, indicating that periods with slow versus rapid disease progression could be distinguished reasonably well based on weather. Results were similar for the 2008 validation data. However, weather variables were not useful in separating periods with either slow or rapid disease progress from those having intermediate progress. Instead, weather variables were most useful when used in a negative-prognosis approach to predict disease progression as being "not rapid" (which includes slow and intermediate periods) or "not slow" (including intermediate and rapid periods). The data set was further analyzed using Classification and Regression Tree (CART) analysis to relate weekly disease progress periods to weather variables. The resulting CART model agreed with the heuristic approach in that temperature variables were more prominent than moisture variables in classifying disease progress periods. With both approaches, satisfactory accuracy was accomplished only with negative-prognoses that classified disease progress periods as not rapid or not slow based on temperature and moisture limits.

Predicting Survival Outcome of Localized Melanoma: An Electronic Prediction Tool Based on the AJCC Melanoma Database

We sought to develop a reliable and reproducible statistical model to predict the survival outcome of patients with localized melanoma. A total of 25,734 patients with localized melanoma from the 2008 American Joint Committee on Cancer (AJCC) Melanoma Database were used for the model development and validation. The predictive model was developed from the model development data set (n = 14,760) contributed by nine major institutions and study groups and was validated on an independent model validation data set (n = 10,974) consisting of patients from a separate melanoma center. Multivariate analyses based on the Cox model were performed for the model development, and the concordance correlation coefficients were calculated to assess the adequacy of the predictive model. Patient characteristics in both data sets were virtually identical, and tumor thickness was the single most important prognostic factor. Other key prognostic factors identified by stratified analyses included ulceration, lesion site, and patient age. Direct comparisons of the predicted 5- and 10-year survival rates calculated from the predictive model and the observed Kaplan-Meier 5- and 10-year survival rates estimated from the validation data set yielded high concordance correlation coefficients of 0.90 and 0.93, respectively. A Web-based electronic prediction tool was also developed ( http://www.melanomaprognosis.org/ ). This is the first predictive model for localized melanoma that was developed based on a very large data set and was successfully validated on an independent data set. The high concordance correlation coefficients demonstrated the accuracy of the predicted model. This predictive model provides a clinically useful tool for making treatment decisions, for assessing patient risk, and for planning and analyzing clinical trials.