Predictive modeling of blood pressure during hemodialysis: a comparison of linear model, random forest, support vector regression, XGBoost, LASSO regression and ensemble method.

BACKGROUND: Intradialytic hypotension (IDH) is commonly occurred and links to higher mortality among patients undergoing hemodialysis (HD). Its early prediction and prevention will dramatically improve the quality of life. However, predicting the occurrence of IDH clinically is not simple. The aims of this study are to develop an intelligent system with capability of predicting blood pressure (BP) during HD, and to further compare different machine learning algorithms for next systolic BP (SBP) prediction. METHODS: This study presented comprehensive comparisons among linear regression model, least absolute shrinkage and selection operator (LASSO), tree-based ensemble machine learning models (random forest [RF] and extreme gradient boosting [XGBoost]), and support vector regression to predict the BP during HD treatment based on 200 and 48 maintenance HD patients containing a total of 7,180 and 2,065 BP records for the training and test dataset, respectively. Ensemble method also was computed to obtain better predictive performance. We compared the developed models based on R2, root mean square error (RMSE) and mean absolute error (MAE). RESULTS: We found that RF (R2=0.95, RMSE=6.64, MAE=4.90) and XGBoost (R2=1.00, RMSE=1.83, MAE=1.29) had comparable predictive performance on the training dataset. However, RF (R2=0.49, RMSE=16.24, MAE=12.14) had more accurate than XGBoost (R2=0.41, RMSE=17.65, MAE=13.47) on testing dataset. Among these models, the ensemble method (R2=0.50, RMSE=16.01, MAE=11.97) had the best performance on testing dataset for next SBP prediction. CONCLUSIONS: We compared five machine learning and an ensemble method for next SBP prediction. Among all studied algorithms, the RF and the ensemble method have the better predictive performance. The prediction models using ensemble method for intradialytic BP profiling may be able to assist the HD staff or physicians in individualized care and prompt intervention for patients' safety and improve care of HD patients.

Influences of Flap Shape and Hinge Angle on Opaque Bubble Layer Formation in Femtosecond Laser-Assisted LASIK Surgery.

PURPOSE: To evaluate the effects of different flap shapes and hinge angles on opaque bubble layer (OBL) formation using a femtosecond laser for flap creation in LASIK surgery. METHODS: This retrospective study evaluated 138 eyes of 73 patients who underwent femtosecond laser-assisted LASIK with a 150-kHz IntraLase femtosecond laser (Abbott Medical Optics, Inc., Santa Ana, CA) for myopic astigmatism and compared differences between different flap shapes and hinge angles on OBL formation. The surgical procedures were videotaped, and the patterns and sizes of the OBLs seen during surgery were analyzed. Preoperative and postoperative data including patient demographics, refractive status, keratometry, central corneal thickness, and intraoperative data (flap size and pocket parameters) were recorded. The eyes were divided into four groups based on the corneal flap shape (elliptical versus round) and hinge angle (50° versus 60°). RESULTS: The preoperative demographic data, mean spherical errors, cylindrical power, and central corneal thickness were not significantly different among the groups. Of the 138 eyes, 107 (77%) developed an OBL covering a mean area of 13.8% ± 12.6% in each case. This area was significantly smaller in the elliptical flap with 60° hinge angle group (P < .05). CONCLUSIONS: An oval-shaped flap with a larger hinge angle tended to result in less OBL formation in femtosecond laser-assisted LASIK. [J Refract Surg. 2017;33(3):178-182.].

The Associations Between Helicobacter pylori Infection, Serum Vitamin D, and Metabolic Syndrome: A Community-Based Study.

The associations between Helicobacter pylori infection, serum vitamin D level, and metabolic syndrome (MS) are controversial. The present community-based study aimed to investigate the effect of H pylori infection and serum vitamin D deficiency on MS development.Individuals from the northeastern region of Taiwan were enrolled in a community-based study from March, 2014 to August, 2015. All participants completed a demographic survey and underwent the urea breath test (UBT) to detect H pylori infection as well as blood tests to determine levels of vitamin D, adiponectin, leptin, and high-sensitivity C-reactive protein. The ATP III criteria for MS were used in this study.A total of 792 men and 1321 women were enrolled. The mean age was 56.4 ±â€Š13.0 years. After adjusting for age and sex, the estimated odds of MS development for a UBT-positive subject were 1.503 (95% confidence interval [CI]: 1.206-1.872, P < 0.001) when compared to a UBT-negative subject. For participants with vitamin D deficiency (<20 ng/mL), the odds of MS development were 1.423 (95% CI: 1.029-1.967, P = 0.033) when compared to those with sufficient vitamin D level (>30 ng/mL). For participants with both H pylori infection and vitamin D deficiency, the odds of MS development were 2.140 (95% CI: 1.348-3.398, P = 0.001) when compared to subjects without H pylori infection and with sufficient vitamin D levels.H pylori infection and vitamin D deficiency could be predictors of MS. For individuals with both H pylori infection and vitamin D deficiency, the odds of MS development were 2.140 when compared to individuals without H pylori infection and with sufficient vitamin D levels.

Helicobacter pylori Infection Increases Insulin Resistance and Metabolic Syndrome in Residents Younger than 50 Years Old: A Community-Based Study.

This study aimed to analyze the influence of H. pylori infection on insulin resistance and metabolic syndrome (MS) by multivariate analysis of a community-based cohort study. From January 2013 to February 2014,811 subjects were enrolled in a community-based cohort study from the northeastern region of Taiwan. All subjects received a demographic survey and blood tests, including an H. pylori antibody test, liver biochemistry tests, lipid profiles, sugar/insulin levels for Homeostatic model assessment (HOMA-IR index), and measurements of adipokines and inflammatory cytokines. A total of 264 men and 547 women were included in this study. The mean age was 59.2 ± 12.7 years. Subjects seropositive for H. pylori antibodies exhibited higher rates of hypertension, an increased incidence of a HOMA-IR index > 2.5 and a higher level of tumor necrosis factor-α than those without H. pylori antibodies. We found a significant difference in the presence of H. pylori antibodies between subjects with MS and those without MS (76.7% vs. 53.7%, p = 0.007) among subjects < 50 y/o. A HOMA-IR index >2.5, H. pylori antibody presence and leptin were predictors for MS in subjects < 50 y/o. The estimated odds ratio of MS for a subject with H. pylori antibodies was 3.717 (95% CI = 1.086-12.719) times that of a subject without H. pylori antibodies. In addition, no difference in H. pylori antibody status was detected for MS prediction in subjects that were ≧ 50 y/o (p = 0.861). In conclusion, subjects with H. pylori antibodies had a higher incidence of a HOMA-IR >2.5 than those without H pylori antibodies. For subjects aged < 50 y/o, the H. pylori antibody was a predictor for MS.

MAVTgsa: an R package for gene set (enrichment) analysis.

Gene set analysis methods aim to determine whether an a priori defined set of genes shows statistically significant difference in expression on either categorical or continuous outcomes. Although many methods for gene set analysis have been proposed, a systematic analysis tool for identification of different types of gene set significance modules has not been developed previously. This work presents an R package, called MAVTgsa, which includes three different methods for integrated gene set enrichment analysis. (1) The one-sided OLS (ordinary least squares) test detects coordinated changes of genes in gene set in one direction, either up- or downregulation. (2) The two-sided MANOVA (multivariate analysis variance) detects changes both up- and downregulation for studying two or more experimental conditions. (3) A random forests-based procedure is to identify gene sets that can accurately predict samples from different experimental conditions or are associated with the continuous phenotypes. MAVTgsa computes the P values and FDR (false discovery rate) q-value for all gene sets in the study. Furthermore, MAVTgsa provides several visualization outputs to support and interpret the enrichment results. This package is available online.

Highly conserved influenza A virus epitope sequences as candidates of H3N2 flu vaccine targets.

This study focused on identifying the conserved epitopes in a single subtype A (H3N2)-as candidates for vaccine targets. We identified a total of 32 conserved epitopes in four viral proteins [22 HA, 4PB1, 3 NA, 3 NP]. Evaluation of conserved epitopes in coverage during 1968-2010 revealed that (1) 12 HA conserved epitopes were highly present in the circulating viruses; (2) the remaining 10 HA conserved epitopes appeared with lower percentage but a significantly increasing trend after 1989 [p<0.001]; and (3) the conserved epitopes in NA, NP and PB1 are also highly frequent in wild-type viruses. These conserved epitopes also covered an extremely high percentage of the 16 vaccine strains during the 42 year period. The identification of highly conserved epitopes using our approach can also be applied to develop broad-spectrum vaccines.

Optimal sampling in retrospective logistic regression via two-stage method.

Case-control sampling is popular in epidemiological research because of its cost and time saving. In a logistic regression model, with limited knowledge on the covariance matrix of the point estimator of the regression coefficients a priori, there exists no fixed sample size analysis. In this study, we propose a two-stage sequential analysis, in which the optimal sample fraction and the required sample size to achieve a predetermined volume of a joint confidence set are estimated in an interim analysis. Additionally required observations are collected in the second stage according to the estimated optimal sample fraction. At the end of the experiment, data from these two stages are combined and analyzed for statistical inference. Simulation studies are conducted to justify the proposed two-stage procedure and an example is presented for illustration. It is found that the proposed two-stage procedure performs adequately in the sense that the resultant joint confidence set has a well-controlled volume and achieves the required coverage probability. Furthermore, the optimal sample fractions among all the selected scenarios are close to one. Hence, the proposed procedure can be simplified by always considering a balance design.