Disease-free survival after hepatic resection in hepatocellular carcinoma patients: a prediction approach using artificial neural network

Background

A database for hepatocellular carcinoma (HCC) patients who had received hepatic resection was used to develop prediction models for 1-, 3- and 5-year disease-free survival based on a set of clinical parameters for this patient group.

Methods

The three prediction models included an artificial neural network (ANN) model, a logistic regression (LR) model, and a decision tree (DT) model. Data for 427, 354 and 297 HCC patients with histories of 1-, 3- and 5-year disease-free survival after hepatic resection, respectively, were extracted from the HCC patient database. From each of the three groups, 80% of the cases (342, 283 and 238 cases of 1-, 3- and 5-year disease-free survival, respectively) were selected to provide training data for the prediction models. The remaining 20% of cases in each group (85, 71 and 59 cases in the three respective groups) were assigned to validation groups for performance comparisons of the three models. Area under receiver operating characteristics curve (AUROC) was used as the performance index for evaluating the three models.

Conclusions

The ANN model outperformed the LR and DT models in terms of prediction accuracy. This study demonstrated the feasibility of using ANNs in medical decision support systems for predicting disease-free survival based on clinical databases in HCC patients who have received hepatic resection.     

Comparisons of Prediction Models of Quality of Life after Laparoscopic Cholecystectomy: A Longitudinal Prospective Study

Background

Few studies of laparoscopic cholecystectomy (LC) outcome have used longitudinal data for more than two years. Moreover, no studies have considered group differences in factors other than outcome such as age and nonsurgical treatment. Additionally, almost all published articles agree that the essential issue of the internal validity (reproducibility) of the artificial neural network (ANN), support vector machine (SVM), Gaussian process regression (GPR) and multiple linear regression (MLR) models has not been adequately addressed. This study proposed to validate the use of these models for predicting quality of life (QOL) after LC and to compare the predictive capability of ANNs with that of SVM, GPR and MLR.

Methodology/Principal Findings

A total of 400 LC patients completed the SF-36 and the Gastrointestinal Quality of Life Index at baseline and at 2 years postoperatively. The criteria for evaluating the accuracy of the system models were mean square error (MSE) and mean absolute percentage error (MAPE). A global sensitivity analysis was also performed to assess the relative significance of input parameters in the system model and to rank the variables in order of importance. Compared to SVM, GPR and MLR models, the ANN model generally had smaller MSE and MAPE values in the training data set and test data set. Most ANN models had MAPE values ranging from 4.20% to 8.60%, and most had high prediction accuracy. The global sensitivity analysis also showed that preoperative functional status was the best parameter for predicting QOL after LC.

Conclusions/Significance

Compared with SVM, GPR and MLR models, the ANN model in this study was more accurate in predicting patient-reported QOL and had higher overall performance indices. Further studies of this model may consider the effect of a more detailed database that includes complications and clinical examination findings as well as more detailed outcome data.     

Comparison of Prediction Model for Cardiovascular Autonomic Dysfunction Using Artificial Neural Network and Logistic Regression Analysis

Background

This study aimed to develop the artificial neural network (ANN) and multivariable logistic regression (LR) analyses for prediction modeling of cardiovascular autonomic (CA) dysfunction in the general population, and compare the prediction models using the two approaches.

Methods and Materials

We analyzed a previous dataset based on a Chinese population sample consisting of 2,092 individuals aged 30–80 years. The prediction models were derived from an exploratory set using ANN and LR analysis, and were tested in the validation set. Performances of these prediction models were then compared.

Results

Univariate analysis indicated that 14 risk factors showed statistically significant association with the prevalence of CA dysfunction (P<0.05). The mean area under the receiver-operating curve was 0.758 (95% CI 0.724–0.793) for LR and 0.762 (95% CI 0.732–0.793) for ANN analysis, but noninferiority result was found (P<0.001). The similar results were found in comparisons of sensitivity, specificity, and predictive values in the prediction models between the LR and ANN analyses.

Conclusion

The prediction models for CA dysfunction were developed using ANN and LR. ANN and LR are two effective tools for developing prediction models based on our dataset.     

A Model to Discriminate Malignant from Benign Thyroid Nodules Using Artificial Neural Network

Objective

This study aimed to construct a model for using in differentiating benign and malignant nodules with the artificial neural network and to increase the objective diagnostic accuracy of US.

Materials and methods

618 consecutive patients (528 women, 161 men) with 689 thyroid nodules (425 malignant and 264 benign nodules) were enrolled in the present study. The presence and absence of each sonographic feature was assessed for each nodule - shape, margin, echogenicity, internal composition, presence of calcifications, peripheral halo and vascularity on color Doppler. The variables meet the following criteria: important sonographic features and statistically significant difference were selected as the input layer to build the ANN for predicting the malignancy of nodules.

Results

Six sonographic features including shape (Taller than wide, p<0.001), margin (Not Well-circumscribed, p<0.001), echogenicity (Hypoechogenicity, p<0.001), internal composition (Solid, p<0.001), presence of calcifications (Microcalcification, p<0.001) and peripheral halo (Absent, p<0.001) were significantly associated with malignant nodules. A three-layer 6-8-1 feed-forward ANN model was built. In the training cohort, the accuracy of the ANN in predicting malignancy of thyroid nodules was 82.3% (AUROC = 0.818), the sensitivity and specificity was 84.5% and 79.1%, respectively. In the validation cohort, the accuracy, sensitivity and specificity was 83.1%, 83.8% and 81.8%, respectively. The AUROC was 0.828.

Conclusion

ANN constructed by sonographic features can discriminate benign and malignant thyroid nodules with high diagnostic accuracy.     

A New Equation to Estimate Glomerular Filtration Rate in Chinese Elderly Population

Background

We sought to develop a new equation to estimate glomerular filtration rate (GFR) in Chinese elderly population.

Methods

A total of 668 Chinese elderly participants, including the development cohort (n = 433), the validation cohort (n = 235) were enrolled. The new equation using the generalized additive model, and age, gender, serum creatinine as predictor variables was developed and the performances was compared with the CKD-EPI equation.

Results

In the validation data set, both bias and precision were improved with the new equation, as compared with the CKD-EPI equation (median difference, −1.5 ml/min/1.73 m² vs. 7.4 ml/min/1.73 m² for the new equation and the CKD-EPI equation, [P<0.001]; interquartile range [IQR] for the difference, 16.2 ml/min/1.73 m² vs. 19.0 ml/min/1.73 m² [P<0.001]), as were accuracies (15% accuracy, 40.4% vs. 30.6% [P = 0.02]; 30% accuracy, 71.1% vs. 47.2%, [P<0.001]; 50% accuracy, 90.2% vs. 75.7%, [P<0.001]), allowing improvement in GFR categorization (GFR category misclassification rate, 37.4% vs. 53.2% [P = <0.001]).

Conclusions

A new equation was developed in Chinese elderly population. In the validation data set, the new equation performed better than the original CKD-EPI equation. The new equation needs further external validations. Calibration of the GFR referent standard to a more accurate one should be an useful way to improve the performance of GFR estimating equations.     

A Digital Network Approach to Infer Sex Behavior in Emerging HIV Epidemics

Purpose

Improve the ability to infer sex behaviors more accurately using network data.

Methods

A hybrid network analytic approach was utilized to integrate: (1) the plurality of reports from others tied to individual(s) of interest; and (2) structural features of the network generated from those ties. Network data was generated from digitally extracted cell-phone contact lists of a purposeful sample of 241 high-risk men in India. These data were integrated with interview responses to describe the corresponding individuals in the contact lists and the ties between them. HIV serostatus was collected for each respondent and served as an internal validation of the model’s predictions of sex behavior.

Results

We found that network-based model predictions of sex behavior and self-reported sex behavior had limited correlation (54% agreement). Additionally, when respondent sex behaviors were re-classified to network model predictions from self-reported data, there was a 30.7% decrease in HIV seroprevalence among groups of men with lower risk behavior, which is consistent with HIV transmission biology.

Conclusion

Combining the relative completeness and objectivity of digital network data with the substantive details of classical interview and HIV biomarker data permitted new analyses and insights into the accuracy of self-reported sex behavior.     

Using Highly Detailed Administrative Data to Predict Pneumonia Mortality

Background

Mortality prediction models generally require clinical data or are derived from information coded at discharge, limiting adjustment for presenting severity of illness in observational studies using administrative data.

Objectives

To develop and validate a mortality prediction model using administrative data available in the first 2 hospital days.

Research Design

After dividing the dataset into derivation and validation sets, we created a hierarchical generalized linear mortality model that included patient demographics, comorbidities, medications, therapies, and diagnostic tests administered in the first 2 hospital days. We then applied the model to the validation set.

Subjects

Patients aged ≥18 years admitted with pneumonia between July 2007 and June 2010 to 347 hospitals in Premier, Inc.’s Perspective database.

Measures

In hospital mortality.

Results

The derivation cohort included 200,870 patients and the validation cohort had 50,037. Mortality was 7.2%. In the multivariable model, 3 demographic factors, 25 comorbidities, 41 medications, 7 diagnostic tests, and 9 treatments were associated with mortality. Factors that were most strongly associated with mortality included receipt of vasopressors, non-invasive ventilation, and bicarbonate. The model had a c-statistic of 0.85 in both cohorts. In the validation cohort, deciles of predicted risk ranged from 0.3% to 34.3% with observed risk over the same deciles from 0.1% to 33.7%.

Conclusions

A mortality model based on detailed administrative data available in the first 2 hospital days had good discrimination and calibration. The model compares favorably to clinically based prediction models and may be useful in observational studies when clinical data are not available.     

A New Modified CKD-EPI Equation for Chinese Patients with Type 2 Diabetes

Objective

To improve the performance of glomerular filtration rate (GFR) estimating equation in Chinese type 2 diabetic patients by modification of the CKD-EPI equation.

Design and patients

A total of 1196 subjects were enrolled. Measured GFR was calibrated to the dual plasma sample ^99mTc-DTPA-GFR. GFRs estimated by the re-expressed 4-variable MDRD equation, the CKD-EPI equation and the Asian modified CKD-EPI equation were compared in 351 diabetic/non-diabetic pairs. And a new modified CKD-EPI equation was reconstructed in a total of 589 type 2 diabetic patients.

Results

In terms of both precision and accuracy, GFR estimating equations all achieved better results in the non-diabetic cohort comparing with those in the type 2 diabetic cohort (30% accuracy, P≤0.01 for all comparisons). In the validation data set, the new modified equation showed less bias (median difference, 2.3 ml/min/1.73 m² for the new modified equation vs. ranged from −3.8 to −7.9 ml/min/1.73 m² for the other 3 equations [P<0.001 for all comparisons]), as was precision (IQR of the difference, 24.5 ml/min/1.73 m² vs. ranged from 27.3 to 30.7 ml/min/1.73 m²), leading to a greater accuracy (30% accuracy, 71.4% vs. 55.2% for the re-expressed 4 variable MDRD equation and 61.0% for the Asian modified CKD-EPI equation [P = 0.001 and P = 0.02]).

Conclusion

A new modified CKD-EPI equation for type 2 diabetic patients was developed and validated. The new modified equation improves the performance of GFR estimation.     

Metabolic Syndrome and Chronic Kidney Disease in an Adult Korean Population: Results from the Korean National Health Screening

Background

This study was aimed to examine the prevalence of metabolic syndrome (MS) and chronic kidney disease (CKD), and the association between MS and its components with CKD in Korea.

Methods

We excluded diabetes to appreciate the real impact of MS and performed a cross-sectional study using the general health screening data of 10,253,085 (48.86±13.83 years, men 56.18%) participants (age, ≥20 years) from the Korean National Health Screening 2011. CKD was defined as dipstick proteinuria ≥1 or an estimated glomerular filtration rate (eGFR) <60 ml/min/1.73 m².

Results

The prevalence of CKD was 6.15% (men, 5.37%; women, 7.15%). Further, 22.25% study population had MS (abdominal obesity, 27.98%; hypertriglyceridemia, 30.09%; low high-density cholesterol levels, 19.74%; high blood pressure, 43.45%; and high fasting glucose levels, 30.44%). Multivariate-adjusted analysis indicated that proteinuria risk increased in participants with MS (odds ratio [OR] 1.884, 95% confidence interval [CI] 1.867–1.902, P<0.001). The presence of MS was associated with eGFR<60 mL/min/1.73 m² (OR 1.364, 95% CI 1.355–1.373, P<0.001). MS individual components were also associated with an increased CKD risk. The strength of association between MS and the development of CKD increase as the number of components increased from 1 to 5. In sub-analysis by men and women, MS and its each components were a significant determinant for CKD.

Conclusions

MS and its individual components can predict the risk of prevalent CKD for men and women.     

Novel Multiple Markers to Distinguish Melanoma from Dysplastic Nevi

Background

Distinguishing melanoma from dysplastic nevi can be challenging.

Objective

To assess which putative molecular biomarkers can be optimally combined to aid in the clinical diagnosis of melanoma from dysplastic nevi.

Methods

Immunohistochemical expressions of 12 promising biomarkers (pAkt, Bim, BRG1, BRMS1, CTHRC1, Cul1, ING4, MCL1, NQO1, SKP2, SNF5 and SOX4) were studied in 122 melanomas and 33 dysplastic nevi on tissue microarrays. The expression difference between melanoma and dysplastic nevi was performed by univariate and multiple logistic regression analysis, diagnostic accuracy of single marker and optimal combinations were performed by receiver operating characteristic (ROC) curve and artificial neural network (ANN) analysis. Classification and regression tree (CART) was used to examine markers simultaneous optimizing the accuracy of melanoma. Ten-fold cross-validation was analyzed for estimating generalization error for classification.

Results

Four (Bim, BRG1, Cul1 and ING4) of 12 markers were significantly differentially expressed in melanoma compared with dysplastic nevi by both univariate and multiple logistic regression analysis (p < 0.01). These four combined markers achieved 94.3% sensitivity, 81.8% specificity and attained 84.3% area under the ROC curve (AUC) and the ANN classified accuracy with training of 83.2% and testing of 81.2% for distinguishing melanoma from dysplastic nevi. The classification trees identified ING4, Cul1 and BRG1 were the most important classification parameters in ranking top-performing biomarkers with cross-validation error of 0.03.

Conclusions

The multiple biomarkers ING4, Cul1, BRG1 and Bim described here can aid in the discrimination of melanoma from dysplastic nevi and provide a new insight to help clinicians recognize melanoma.     

Circulating Adipocytokines and Chronic Kidney Disease

Background

Adipokines have been associated with atherosclerotic heart disease, which shares many common risk factors with chronic kidney disease (CKD), but their relationship with CKD has not been well characterized.

Methods

We investigated the association of plasma leptin, resistin and adiponectin with CKD in 201 patients with CKD and 201 controls without. CKD was defined as estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m² or presence of albuminuria. Quantile regression and logistic regression models were used to examine the association between adipokines and CKD adjusting for multiple confounding factors.

Results

Compared to controls, adjusted median leptin (38.2 vs. 17.2 ng/mL, p<0.0001) and adjusted mean resistin (16.2 vs 9.0 ng/mL, p<0.0001) were significantly higher in CKD cases. The multiple-adjusted odds ratio (95% confidence interval) of CKD comparing the highest tertile to the lower two tertiles was 2.3 (1.1, 4.9) for leptin and 12.7 (6.5, 24.6) for resistin. Median adiponectin was not significantly different in cases and controls, but the odds ratio comparing the highest tertile to the lower two tertiles was significant (1.9; 95% CI, 1.1, 3.6). In addition, higher leptin, resistin, and adiponectin were independently associated with lower eGFR and higher urinary albumin levels.

Conclusions

These findings suggest that adipocytokines are independently and significantly associated with the risk and severity of CKD. Longitudinal studies are warranted to evaluate the prospective relationship of adipocytokines to the development and progression of CKD.     

Imputation of sequence level genotypes in the Franches-Montagnes horse breed

Background

A cost-effective strategy to increase the density of available markers within a population is to sequence a small proportion of the population and impute whole-genome sequence data for the remaining population. Increased densities of typed markers are advantageous for genome-wide association studies (GWAS) and genomic predictions.

Methods

We obtained genotypes for 54 602 SNPs (single nucleotide polymorphisms) in 1077 Franches-Montagnes (FM) horses and Illumina paired-end whole-genome sequencing data for 30 FM horses and 14 Warmblood horses. After variant calling, the sequence-derived SNP genotypes (~13 million SNPs) were used for genotype imputation with the software programs Beagle, Impute2 and FImpute.

Results

The mean imputation accuracy of FM horses using Impute2 was 92.0%. Imputation accuracy using Beagle and FImpute was 74.3% and 77.2%, respectively. In addition, for Impute2 we determined the imputation accuracy of all individual horses in the validation population, which ranged from 85.7% to 99.8%. The subsequent inclusion of Warmblood sequence data further increased the correlation between true and imputed genotypes for most horses, especially for horses with a high level of admixture. The final imputation accuracy of the horses ranged from 91.2% to 99.5%.

Conclusions

Using Impute2, the imputation accuracy was higher than 91% for all horses in the validation population, which indicates that direct imputation of 50k SNP-chip data to sequence level genotypes is feasible in the FM population. The individual imputation accuracy depended mainly on the applied software and the level of admixture.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-014-0063-7) contains supplementary material, which is available to authorized users.     

Effects of Clopidogrel on Mortality,Cardiovascular and Bleeding Outcomes in Patients with Chronic Kidney Disease - Data from Taiwan Acute Coronary Syndrome Full Spectrum Registry

Background

The efficacy of clopidogrel is inconclusive in the chronic kidney disease (CKD) population with acute coronary syndrome (ACS). Furthermore, CKD patients are prone to bleeding with antiplatelet therapy. We investigated the efficacy and safety of clopidogrel in patients with ACS and CKD.

Methods

In a Taiwan national-wide registry, 2819 ACS patients were enrolled. CKD is defined as an estimated glomerular filtration rate of less than 60 ml/min per 1.73 m². The primary endpoints are the combined outcomes of death, non-fatal myocardial infarction and stroke at 12 months.

Results

Overall 949 (33.7%) patients had CKD and 2660 (94.36%) patients received clopidogrel treatment. CKD is associated with increased risk of the primary endpoint at 12 months (HR 2.39, 95% CI 1.82 to 3.15, p<0.01). Clopidogrel use is associated with reduced risk of the primary endpoint at 12 months (HR 0.42, 95% CI: 0.29–0.60, p<0.01). Cox regression analysis showed that clopidogrel reduced death and primary endpoints for CKD population (HR 0.35, 95% CI: 0.21–0.61 and HR 0.48, 95% CI: 0.30–0.77, respectively, both p<0.01). Patients with clopidogrel(−)/CKD(−), clopidogrel(+)/CKD(+) and clopidogrel(−)/CKD(+) have 2.4, 3.0 and 10.4 fold risk to have primary endpoints compared with those receiving clopidogrel treatment without CKD (all p<0.01). Clopidogrel treatment was not associated with increased in-hospital Thrombolysis In Myocardial Infarction (TIMI) bleeding in CKD population.

Conclusion

Clopidogrel could decrease mortality and improve cardiovascular outcomes without increasing risk of bleeding in ACS patients with CKD.     

Laboratory Test Surveillance following Acute Kidney Injury

Background

Patients with hospitalized acute kidney injury (AKI) are at increased risk for accelerated loss of kidney function, morbidity, and mortality. We sought to inform efforts at improving post-AKI outcomes by describing the receipt of renal-specific laboratory test surveillance among a large high-risk cohort.

Methods

We acquired clinical data from the Electronic health record (EHR) of 5 Veterans Affairs (VA) hospitals to identify patients hospitalized with AKI from January 1^st, 2002 to December 31^st, 2009, and followed these patients for 1 year or until death, enrollment in palliative care, or improvement in renal function to estimated GFR (eGFR) ≥60 L/min/1.73 m². Using demographic data, administrative codes, and laboratory test data, we evaluated the receipt and timing of outpatient testing for serum concentrations of creatinine and any as well as quantitative proteinuria recommended for CKD risk stratification. Additionally, we reported the rate of phosphorus and parathyroid hormone (PTH) monitoring recommended for chronic kidney disease (CKD) patients.

Results

A total of 10,955 patients admitted with AKI were discharged with an eGFR<60 mL/min/1.73 m². During outpatient follow-up at 90 and 365 days, respectively, creatinine was measured on 69% and 85% of patients, quantitative proteinuria was measured on 6% and 12% of patients, PTH or phosphorus was measured on 10% and 15% of patients.

Conclusions

Measurement of creatinine was common among all patients following AKI. However, patients with AKI were infrequently monitored with assessments of quantitative proteinuria or mineral metabolism disorder, even for patients with baseline kidney disease.     

Comparison of artificial neural network and logistic regression models for predicting in-hospital mortality after primary liver cancer surgery

Background

Since most published articles comparing the performance of artificial neural network (ANN) models and logistic regression (LR) models for predicting hepatocellular carcinoma (HCC) outcomes used only a single dataset, the essential issue of internal validity (reproducibility) of the models has not been addressed. The study purposes to validate the use of ANN model for predicting in-hospital mortality in HCC surgery patients in Taiwan and to compare the predictive accuracy of ANN with that of LR model.

Methodology/Principal Findings

Patients who underwent a HCC surgery during the period from 1998 to 2009 were included in the study. This study retrospectively compared 1,000 pairs of LR and ANN models based on initial clinical data for 22,926 HCC surgery patients. For each pair of ANN and LR models, the area under the receiver operating characteristic (AUROC) curves, Hosmer-Lemeshow (H-L) statistics and accuracy rate were calculated and compared using paired T-tests. A global sensitivity analysis was also performed to assess the relative significance of input parameters in the system model and the relative importance of variables. Compared to the LR models, the ANN models had a better accuracy rate in 97.28% of cases, a better H-L statistic in 41.18% of cases, and a better AUROC curve in 84.67% of cases. Surgeon volume was the most influential (sensitive) parameter affecting in-hospital mortality followed by age and lengths of stay.

Conclusions/Significance

In comparison with the conventional LR model, the ANN model in the study was more accurate in predicting in-hospital mortality and had higher overall performance indices. Further studies of this model may consider the effect of a more detailed database that includes complications and clinical examination findings as well as more detailed outcome data.     

Using Mathematical Algorithms to Modify Glomerular Filtration Rate Estimation Equations

Background

The equations provide a rapid and low-cost method of evaluating glomerular filtration rate (GFR). Previous studies indicated that the Modification of Diet in Renal Disease (MDRD), Chronic Kidney Disease-Epidemiology (CKD-EPI) and MacIsaac equations need further modification for application in Chinese population. Thus, this study was designed to modify the three equations, and compare the diagnostic accuracy of the equations modified before and after.

Methodology

With the use of ^{99 m}Tc-DTPA renal dynamic imaging as the reference GFR (rGFR), the MDRD, CKD-EPI and MacIsaac equations were modified by two mathematical algorithms: the hill-climbing and the simulated-annealing algorithms.

Results

A total of 703 Chinese subjects were recruited, with the average rGFR 77.14±25.93 ml/min. The entire modification process was based on a random sample of 80% of subjects in each GFR level as a training sample set, the rest of 20% of subjects as a validation sample set. After modification, the three equations performed significant improvement in slop, intercept, correlated coefficient, root mean square error (RMSE), total deviation index (TDI), and the proportion of estimated GFR (eGFR) within 10% and 30% deviation of rGFR (P₁₀ and P₃₀). Of the three modified equations, the modified CKD-EPI equation showed the best accuracy.

Conclusions

Mathematical algorithms could be a considerable tool to modify the GFR equations. Accuracy of all the three modified equations was significantly improved in which the modified CKD-EPI equation could be the optimal one.     

Comparing Artificial Neural Networks,General Linear Models and Support Vector Machines in Building Predictive Models for Small Interfering RNAs

Background

Exogenous short interfering RNAs (siRNAs) induce a gene knockdown effect in cells by interacting with naturally occurring RNA processing machinery. However not all siRNAs induce this effect equally. Several heterogeneous kinds of machine learning techniques and feature sets have been applied to modeling siRNAs and their abilities to induce knockdown. There is some growing agreement to which techniques produce maximally predictive models and yet there is little consensus for methods to compare among predictive models. Also, there are few comparative studies that address what the effect of choosing learning technique, feature set or cross validation approach has on finding and discriminating among predictive models.

Principal Findings

Three learning techniques were used to develop predictive models for effective siRNA sequences including Artificial Neural Networks (ANNs), General Linear Models (GLMs) and Support Vector Machines (SVMs). Five feature mapping methods were also used to generate models of siRNA activities. The 2 factors of learning technique and feature mapping were evaluated by complete 3×5 factorial ANOVA. Overall, both learning techniques and feature mapping contributed significantly to the observed variance in predictive models, but to differing degrees for precision and accuracy as well as across different kinds and levels of model cross-validation.

Conclusions

The methods presented here provide a robust statistical framework to compare among models developed under distinct learning techniques and feature sets for siRNAs. Further comparisons among current or future modeling approaches should apply these or other suitable statistically equivalent methods to critically evaluate the performance of proposed models. ANN and GLM techniques tend to be more sensitive to the inclusion of noisy features, but the SVM technique is more robust under large numbers of features for measures of model precision and accuracy. Features found to result in maximally predictive models are not consistent across learning techniques, suggesting care should be taken in the interpretation of feature relevance. In the models developed here, there are statistically differentiable combinations of learning techniques and feature mapping methods where the SVM technique under a specific combination of features significantly outperforms all the best combinations of features within the ANN and GLM techniques.     

Neural Network Prediction of ICU Length of Stay Following Cardiac Surgery Based on Pre-Incision Variables

Background

Advanced predictive analytical techniques are being increasingly applied to clinical risk assessment. This study compared a neural network model to several other models in predicting the length of stay (LOS) in the cardiac surgical intensive care unit (ICU) based on pre-incision patient characteristics.

Methods

Thirty six variables collected from 185 cardiac surgical patients were analyzed for contribution to ICU LOS. The Automatic Linear Modeling (ALM) module of IBM-SPSS software identified 8 factors with statistically significant associations with ICU LOS; these factors were also analyzed with the Artificial Neural Network (ANN) module of the same software. The weighted contributions of each factor (“trained” data) were then applied to data for a “new” patient to predict ICU LOS for that individual.

Results

Factors identified in the ALM model were: use of an intra-aortic balloon pump; O₂ delivery index; age; use of positive cardiac inotropic agents; hematocrit; serum creatinine ≥ 1.3 mg/deciliter; gender; arterial pCO₂. The r² value for ALM prediction of ICU LOS in the initial (training) model was 0.356, p <0.0001. Cross validation in prediction of a “new” patient yielded r² = 0.200, p <0.0001. The same 8 factors analyzed with ANN yielded a training prediction r² of 0.535 (p <0.0001) and a cross validation prediction r² of 0.410, p <0.0001. Two additional predictive algorithms were studied, but they had lower prediction accuracies. Our validated neural network model identified the upper quartile of ICU LOS with an odds ratio of 9.8(p <0.0001).

Conclusions

ANN demonstrated a 2-fold greater accuracy than ALM in prediction of observed ICU LOS. This greater accuracy would be presumed to result from the capacity of ANN to capture nonlinear effects and higher order interactions. Predictive modeling may be of value in early anticipation of risks of post-operative morbidity and utilization of ICU facilities.     

Chronic Kidney Disease Is Characterized by “Double Trouble” Higher Pulse Pressure plus Night-Time Systolic Blood Pressure and More Severe Cardiac Damage

Background

Hypertension plays a key role in chronic kidney disease (CKD), but CKD itself affects the blood pressure (BP) profile. The aim of this study was to assess the association of BP profile with CKD and the presence of cardiac organ damage.

Methods

We studied 1805 patients, referred to our Hypertension Centre, in whom ABPM, blood tests, and echocardiography were clinically indicated. The glomerular filtration rate was estimated (eGFR) using the MDRD equation and CKD was defined as eGFR<60 mL/min/1.73 m². Cardiac organ damage was evaluated by echocardiography.

Results

Among patients with CKD there were higher systolic blood pressure (SBP) during the night-time, greater prevalence of non-dippers (OR: 1.8) and increased pulse pressure (PP) during 24-hour period, daytime and night-time (all p<0.001). Patients with CKD had a greater LVM/h^2.7 index, and a higher prevalence of left ventricular hypertrophy and diastolic dysfunction (all p<0.001). Nocturnal SBP and PP correlated more strongly with cardiac organ damage (p<0.001). Patients with CKD had a greater Treatment Intensity Score (p<0.001) in the absence of a significantly greater BP control.

Conclusions

CKD patients have an altered night-time pressure profile and higher PP that translate into a more severe cardiac organ damage. In spite of a greater intensity of treatment in most patients with CKD, BP control was similar to patients without CKD. Our findings indicate the need of a better antihypertensive therapy in CKD, better selected drugs, dosages and posology to provide optimal coverage of 24 hours and night-time BP.     

Angiotensin-Converting Enzyme Insertion/Deletion Polymorphism Contributes High Risk for Chronic Kidney Disease in Asian Male with Hypertension–A Meta-Regression Analysis of 98 Observational Studies

Background

Associations between angiotensin-converting enzyme (ACE) gene insertion/deletion (I/D) polymorphisms and chronic kidney disease (CKD) have been extensively studied, with most studies reporting that individuals with the D allele have a higher risk. Although some factors, such as ethnicity, may moderate the association between ACE I/D polymorphisms and CKD risk, gender-dependent effects on the CKD risk remain controversial.

Objectives

This study investigated the gender-dependent effects of ACE I/D polymorphisms on CKD risk.

Data sources

PubMed, the Cochrane library, and EMBASE were searched for studies published before January 2013.

Study eligibility criteria, participants, and interventions

Cross-sectional surveys and case–control studies analyzing ACE I/D polymorphisms and CKD were included. They were required to match the following criteria: age >18 years, absence of rare diseases, and Asian or Caucasian ethnicity.

Study appraisal and synthesis methods

The effect of carrying the D allele on CKD risk was assessed by meta-analysis and meta-regression using random-effects models.

Results

Ethnicity [odds ratio (OR): 1.24; 95% confidence interval (CI): 1.08–1.42] and hypertension (OR: 1.55; 95% CI: 1.04–2.32) had significant moderate effects on the association between ACE I/D polymorphisms and CKD risk, but they were not significant in the diabetic nephropathy subgroup. Males had higher OR for the association between ACE I/D polymorphisms and CKD risk than females in Asians but not Caucasians, regardless of adjustment for hypertension (p<0.05). In subgroup analyses, this result was significant in the nondiabetic nephropathy group. Compared with the I allele, the D allele had the highest risk (OR: 3.75; 95% CI: 1.84–7.65) for CKD in hypertensive Asian males.

Conclusions and implications of key findings

The ACE I/D polymorphisms may incur the highest risk for increasing CKD in hypertensive Asian males.