Day 2 :
Keynote Forum
Mikhail Moshkov
King Abdullah University of Science and Technology
Saudi Arabia
Keynote: Extensions of dynamic programming for decision tree study
Time : 09:00-09:30
Biography:
Mikhail Moshkov is Professor in the CEMSE Division at King Abdullah University of Science and Technology, Saudi Arabia since October 1, 2008. He earned Master’s degree from Nizhni Novgorod State University, received his Doctorate from Saratov State University, and Habilitation from Moscow State University. From 1977 to 2004, he was with Nizhni Novgorod State University. Since 2003, he worked in Poland in the Institute of Computer Science, University of Silesia, and since 2006, also in the Katowice Institute of Information Technologies. His main areas of research are complexity of algorithms, combinatorial optimization, and machine learning. He is Author or Co-Author of five research monographs published by Springer.
Abstract:
In the presentation, we consider extensions of dynamic programming approach to the study of decision trees as algorithms for problem solving, as a way for knowledge extraction and representation, and as classifiers which, for a new object given by values of conditional attributes, define a value of the decision attribute. These extensions allow us (i) to describe the set of optimal decision trees, (ii) to count the number of these trees, (iii) to make sequential optimization of decision trees relative to different criteria, (iv) to find the set of Pareto optimal points for two criteria, and (v) to describe relationships between two criteria. The results include the minimization of average depth for decision trees sorting eight elements (this question was open since 1968), improvement of upper bounds on the depth of decision trees for diagnosis of 0-1-faults in read-once combinatorial circuits, existence of totally optimal (with minimum depth and minimum number of nodes) decision trees for monotone Boolean functions with at most six variables, study of time-memory tradeoff for decision trees for corner point detection, study of relationships between number and maximum length of decision rules derived from decision trees, study of accuracy-size tradeoff for decision trees which allows us to construct enough small and accurate decision trees for knowledge representation, and decision trees that, as classifiers, outperform often decision trees constructed by CART. The end of the presentation is devoted to the introduction to KAUST.
Keynote Forum
Joel E Michalek
University of Texas
USA
Keynote: Agent orange exposure and monoclonal gammopathy of undetermined Significance: A ranch hand veteran cohort study
Time : 09:30-10:00
Biography:
Joel Michalek completed his PhD at the age of 29 years from Wayne State University. I has a broad background in biostatistics pertaining to theory and methods, preclinical and clinical trials, and epidemiology. He has written protocols and grants, analyzed data, and co-authored manuscripts arising from clinical studies in surgery, emergency medicine, cancer, and pediatrics and was formerly Principal Investigator of the Air Force Health Study, a 20-year prospective epidemiological study of veterans who sprayed Agent Orange and other herbicides in Vietnam. He has authored 180 journal articles and two book chapters.
Abstract:
Multiple myeloma has been classified as exhibiting “limited or suggestive evidence” of an association with exposure to herbicides in Vietnam Veterans. Occupational studies have shown that other pesticides (i.e., insecticides, herbicides, fungicides) are associated with excess risk of multiple myeloma (MM) and its precursor state, monoclonal gammopathy of undetermined significance (MGUS); however no studies have uncovered such an association in Vietnam Veterans. Our objective was to examine the relationship between MGUS and exposure to Agent Orange, including its contaminant 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), in Vietnam Veterans. We conducted a prospective cohort study, testing for MGUS in serum specimens collected and stored in 2002 by the Air Force Health Study (AFHS). The relevant exposure data collected by the AFHS was also used. We tested all specimens in 2013 without knowledge of the exposure status. The AFHS included former US Air Force personnel who participated in Operation Ranch Hand (Ranch Hand Veterans) and other US Air Force personnel who had similar duties in Southeast Asia during the same time period but were not involved in herbicide spray missions (comparison Veterans). We included 479 Ranch Hand Veterans and 479 comparison veterans who participated in the 2002 follow-up examination of AFHS. Agent Orange and 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD, a contaminant of Agent Orange) measured on a lipid weight basis in serum in 1985, 1987, 1992, 1997, and 2002. The prevalence of MGUS in Ranch Hand Veterans (7.1%) was higher than in comparison Veterans (3.1%) (adjusted OR=2.37, 95% CI, 1.27-4.44; P=0.007). The cohort status was significantly (P=0.0001) associated with TCDD levels: 47% of Ranch Hand Veterans had serum TCDD levels >10.92 ppt compared to 2.5% of comparison Veterans. Ranch Hand Veterans have a significantly increased risk of MGUS, supporting an association between Agent Orange exposure and multiple myeloma.
Keynote Forum
Ahmed A Bahnassy
King Fahad Medical City, Saudi Arabia
Keynote: Factor analysis technique to determine Job Satisfaction among physicians in a Saudi Tertiary Medical Care Center
Time : 10:00-10:30
Biography:
Ahmed A Bahnassy has completed his PhD from Tulane University. He is a Professor of Biostatistics and Medical Research in College of Medicine, King Fahad Medical City, Saudi Arabia. He has published more than 130 papers in reputed journals and has been serving as a Referee of 6 repute journals.
Abstract:
Background:rnPhysician satisfaction decreased over years due to the amount of time spent with individual patients, personal autonomy, and time available for family and personal life. rnThis study attempts to address physician satisfaction, correlates and predictors of high and low satisfaction levelsrnObjectives:rnTo measure the degree of job satisfaction among physicians working in a tertiary care hospital, to identify factors might affect job satisfaction among physicians.rnMaterial and Method:rnThis is a cross section study for 340 physicians selected from a tertiary care center using stratified random sample with proportional allocation using self-administered questionnaire with 5 points Likert Scale. Only 217 completed the questionnaire. Descriptive statistics was used appropriately, Mean + standard deviation for the quantitative variables while frequency and percentages for the qualitative variables. ANOVA, t-test, and Chi-square were used as necessary to find if there are any significant relationships between satisfaction scores and the predictor variables. Factor Analysis technique was used to find the hidden factors for job satisfaction among the surveyed physicians. Scree plot and Eigen values greater than 1 used to determine number of factors.rnResults:rnResponse rate was 63.8% from all physicians selected to participate in the study. They were mostly males (75.6%), non-Saudis (52.5%). The overall perceived satisfaction as measured by one question was 3.42 points out of 5 (68.4%) significantly lower than the overall satisfaction which took in consideration all variables 3.67 points (73.4%). Mean satisfaction scores were significantly negatively related to number of children (p<0.001), but positively correlated to each of: income, amount of vacations, sick leave policy, health coverage for the employee and his family, overall benefits package, involving in academic work, and doing research, (p<0.001).rnConclusion:rn Work culture, relation with supervisors, benefits, appreciation , environment, encouragements, team work, development and privileges are the most important factors affect physician’ satisfaction Boosting satisfaction of physicians is important for both the success of the tertiary care center and for the high quality services offered to the patientsrnKey Words: rnFactor Analysis, Tertiary care centers, Job satisfaction, Physicians, Saudi Arabiarn
- Track 1: Medical and Clinical Biostatistics; Track 3: Bayesian Statistics; Track 4: Regression Analysis
Session Introduction
Aiguo Li
National Institutes of Health
USA
Title: Glioma classification and translational application in clinics
Time : 10:50-11:10
Biography:
Aiguo Li is a Senior Bioinformatician in Neuro-Oncology Branch, National Cancer Institute, National Institutes of Health. Her research interests include glioma tumor biomarker identification and prognostic studies, and tumor molecular and functional classification. The goals of her research are to understand the underlying molecular mechanisms of glioma tumorigenesis and its progression; and developing novel therapeutic approaches for curing glioma patients. In the past, she has established a glioma molecular classification model containing six sub-types and further translated this model into a clinically useful tool – GliomaPredict, which allows clinicians and researchers to assign new patients to the existing sub-types. The focus of her research is, currently, on understanding the glioma tumor’s heterogeneity and infiltration mechanism by integrative analysis of multi-dimensional high throughput data of glioma patients and glioma stem cell lines.
Abstract:
Gliomas are the most common type of primary brain tumors in adults and a significant cause of cancer-related mortality. Defining glioma sub-types based on objective genetic and molecular signatures may allow for a more rational, patient specific approach to therapy in the future. Applying two unsupervised machine-learning methods to 159 glioma patient gene expression profiles ranging from low to high grade gliomas, we have established a glioma classification model containing six distinct sub-types. These sub-types are validated using three additional data-sets and annotated for underlying molecular functions. To translate this glioma classification model into clinical application, we developed a web-based tool, GliomaPredict, for assigning new patients into our molecular sub-types. The classification model also facilitates us to study glioma progression mechanism in cohered and to design targeted clinical trials.
Huaping Wang
University of Illinois at Chicago
USA
Title: Using different methods to analyze the pre-post no-matching survey data in medical practical research
Time : 11:10-11:30
Biography:
Huaping Wang has completed her PhD in Statistics from University of Alabama in 2007. She is a Research Assistant Professor of the University Of Illinois College Of Medicine at Peoria Division of Research Services. She has worked with various projects from many different medical fields, such as pediatrics, medicine, emergency medicine, neurology, neurosurgery, surgery, OB/GYN, and radiology. She has many years of experience providing statistical consulting to the UIC faculty, residents, MD students, and local medical research investigators.
Abstract:
Some problems arise when analyzing pre and post survey studies using anonymous questionnaire. The pre and post data is related, but it is impossible to match the pre and post data of the same responders. Teen drivers have higher crash risk when they drive after consuming alcohol. The American Red Cross Central Illinois Chapter conducts “Operation Prom Night” crash re-enactment programs with local high school students and the Illinois Department of Transportation. The goal is to teach students about the danger of drinking and driving. The survey objective is to measure changes in knowledge and behavior related with drinking and driving. In this situation, some students answered both pre and post survey, some students only answered pre survey or post survey. Pre and post data could not be paired for the same student. The data used was from 13 schools between 2007 and 2011. We used independent sample statistical models to analyze the data in two different ways. One chose all data and another one chose part of the data but independently. In order to create independent samples, we randomly chose 6 out of the 13 schools as group1, and others as group2. The pre or post data was used from group1. By contrast, the post or pre data was used from group2. Both methods decreased the power. The first one was because of extra variability introduced by not knowing the individual, and the second one was because of sample size reduced. All results were compared and had the similar conclusions.
Ahmed A Bahnassy
King Fahad Medical City College of Medicine
Saudi Arabia
Title: Teaching biostatistics to undergraduate Saudi medical students
Time : 10:00-10:30
Biography:
Ahmed A Bahnassy has completed his PhD from Tulane University. He is a Professor of Biostatistics and Medical Research in College of Medicine, King Fahad Medical City, Saudi Arabia. He has published more than 130 papers in reputed journals and has been serving as a Referee of 6 repute journals.
Abstract:
Background: The knowledge and ability to use biostatistical techniques have become increasingly important in management of data and interpretation of results of different of studies related to health sciences. Understanding biostatistical methods may improve clinical thinking, decision making, evaluations, medical research and evidence based healthcare. Unfortunately, there is an increasing gap between medical students and mathematical notations usually part-and-parcel of medical and biostatistics. Lack of connection between medical curricula and introductory courses in statistics has negative attitude among medical students. This study is aimed at evaluating the effect of a new teaching method (which is based on a mix of theoretical concepts and its applications using computer facilities) in teaching biostatistics to undergraduate medical students in a Saudi Faculty of Medicine, and to measure the students’ ability to understand the results of the statistical tests of the computer output and interpret them in meaningful texts. Methods: A new method of teaching biostatistics based on teaching theoretical concepts and application of the lectures using a real clinical dataset using PC-SPSS software. The results of this new teaching method were compared to the conventional method (based on lectures and scientific calculations) in two classes in two academic years. Results: Students in the two classes were 114 students; each class had 57 students. The new method’s class scored significantly higher than those who were taught using the traditional method in the following topics: measures of variability, confidence intervals testing hypothesis, t-test, Analysis of Variance (ANOVA), multiple linear regression and data presentations. Writing and interpreting the results showed borderline statistical difference between the two methods. Mean satisfaction scores for the students toward biostatistics were significantly higher for the new method than the traditional one (p<0.05). Conclusion: Teaching biostatistics for medical students should avoid calculations. Using computer software is recommended for analysis of real medical data.
Ram Shanmugam
Texas State University, USA
Title: Biostatistics can help to comprehend the shortage, illegal trade, and unmet demand of organ or tissue transplant
Time : 13:50-14:10
Biography:
Ram Shanmugam is the Editor-in-Chief for the journals: Epidemiology & Community Medicine, Advances in Life Sciences and Health, and Global Journal of Research and Review. He is the Associate Editor of the International Journal of Research in Medical Sciences. He is the Book-Review Editor of the Journal of Statistical Computation and Simulation. He directed Statistics Consulting Center in the Mississippi State University. In 2015, he has published a textbook with the title “Statistics for Engineers and Scientists”. He served the Argonne National Lab., University of Colorado, University of South Alabama and the Indian Statistical Institute. He has published 125 research articles and is a Fellow of the International Statistical Institute. Currently, he is a Professor in the School of Health Administration, Texas State University. He is a recipient of several research awards from the Texas State University.
Abstract:
Public, in general, and healthcare professionals, in particular, are confused from unclear and conflicting information in the organ transplant related data. To sort out and clarify such confusions, a statistical methodology is constructed and demonstrated in this article. The gap between the number of organ donors and the number waiting for organ transplant is named shortage. The gap between the number of organ donors and the number of recipients is named illegal organ trade level. The gap between the number of organ recipients and the number waiting for organ transplant is named unmet organ demand. Expressions are derived, based on a statistical methodology, to compute the confidence interval for these true unknown gaps. A few recommendations are compiled and stated in the end to close such gaps for the sake of those waiting for organ transplant to have a quality life.
Walid Sharabati
Purdue University
USA
Title: A regression approach for noise removal in image analysis
Time : 14:10-14:30
Biography:
Walid Sharabati joined the Department of Statistics at Purdue University in the fall of 2008. He earned his PhD in Computational Statistics from George Mason University and has an MS in Mathematics and Computer Science from Minnesota State University. His main research interests are social networks, preferential attachment, text mining, stochastic processes, statistical modeling, Gaussian mixture models, and statistical models for image de-blurring and de-noising. He has been serving as an Editorial Board Member of Austin Statistics and Enliven: Biostatistics and Metrics.
Abstract:
Recording an image that is sharp and clear is sometimes challenging, and perturbations are inevitable. Brain CAT scans, for example, may contain noisy regions, and ultrasound images may have unclear object. Galactic images may be blurred and noised because the light is bent by the time it reached the camera or because the existence of dust in the outer space. The objective of image de-noising is to reduce the noise generated during the process of capturing an image, which is due to several factors such as bad camera sensor, memory location or noise in the transmission channel. In this talk, I present a novel approach to tackle Gaussian noise introduced to an image at different levels using a multiple regression model in which the neighboring pixels are the predictors used to estimate the pixel value (color) on the grid. To this end, I consider balls of varying radius values around the predicted pixel. The underlying algorithm portrays a typical inverse problem that requires the introduction of a regularization term to the system. Finally, I utilize the structural similarity index (SSIM) for images and peak signal to noise ratio (PSNR) measure to assess the performance of the model at different noise levels and radii. The results are promising and produce high similarity measure between the de-noised and original sharp images.
Ali Seifi
University of Texas Health Science Center
USA
Title: Healthcare cost and utilization project (H-CUP), a research tool
Time : 14:30-14:50
Biography:
Ali Seifi,MD ,FACP is an assistant professor in the Department of Neurosurgery at the University of Texas Health Science Center San Antonio who has served as an attending physician since 2012. He oversees the care of patients in University Hospitals state-of-the-art Neuro Intensive Care Unit (NICU), as well as leads the units daily operations. This NICU is the only intensive care unit in the region fully dedicated to the care of neurologically ill patients and staffed with physicians and nurses specially trained in the care of patients with brain, spine, and nervous system diseases.
Abstract:
HCUP includes the LARGEST collection of multi-year hospital care (inpatient, outpatient, and emergency department) data in the United States, with all-payer, encounter-level information beginning in 1988. HCUP is the Nation’s most comprehensive source of hospital data, including information on in-patient care, ambulatory care, and emergency department visits. HCUP enables researchers, insurers, policymakers and others to study health care delivery and patient outcomes over time, and at the national, regional, State, and community levels. The Healthcare Cost and Utilization Project (HCUP, pronounced "H-Cup") is a family of databases and related software tools and products developed through a Federal-State-Industry partnership and sponsored by AHRQ. HCUP databases are derived from administrative data and contain encounter-level, clinical and nonclinical information including all-listed diagnoses and procedures, discharge status, patient demographics, and charges for all patients, regardless of payer (e.g., Medicare, Medicaid, private insurance, uninsured), beginning in 1988. These databases enable research on a broad range of health policy issues, including cost and quality of health services, medical practice patterns, access to health care programs, and outcomes of treatments at the national, State, and local market levels. The HCUP databases are based on the data collection efforts of organizations in participating States that maintain statewide data systems and are Partners with AHRQ. In this presentation, we will discuss how to make a clinical research project based on this large database.
Yousri Slaoui
University of Poitiers
France
Title: Parameter estimation in a hierarchical random intercept model with censored response: An approach using a SEM algorithm and Gibbs sampling
Biography:
Yousri Slaoui has completed his PhD from the University of Versailles and Postdoctoral studies from the National Scientific Research Centre of Statistics. He is an Associate Professor at the University of Poitiers. He has published more than 12 papers in reputed journals.
Abstract:
In this talk, we propose an approach, based on the stochastic expectation maximization (SEM) algorithm and Gibbs sampling, to deal with the problem caused by censoring in the response of a hierarchical random intercept models. As an application, we consider a dataset consisting of 2941 parasite density measurement gathered over a population of 505 Senegal children between 2001 and 2003. Assuming that all these measurement are corrects, we simulate the effect of various censure levels by removing the corresponding entries before performing our algorithm. The model residuals are then compared to those obtained with the full data. Even when 10%, 20% or even 30% of the original measurements are missing, the produced residuals remain very accurate thus demontrating the effectiveness of our approach. Moreover, we compared our approach with the existing methods via real data sets as well as simulations. Results showed that our approach outperformed other approaches in terms of estimation accuracy and computing efficiency.
Abbas F Jawad
University of Pennsylvania
Children’s Hospital of Philadelphia
USA
Title: Cluster randomization trials in schools setting
Biography:
Abbas F Jawad has earned his MSc (1986) and PhD (1993) from the Graduate School of Public Health, University of Pittsburgh, USA. He is an Associate Professor of Biostatistics in Pediatrics at the University of Pennsylvania, Perelman School of Medicine, and Department of Pediatrics at the Children’s Hospital of Philadelphia. He has published more than 100 papers in reputed journals and has been providing biostatistical support for medical pediatric research for more than 20 years.
Abstract:
Cluster Randomization Trials in School Setting (CRTSS) are increasingly utilized in studying new and/or existing behavioral and drug therapies targeting students within in school setting. The units of the analyses are the students attending schools but the randomization to the interventions often done at the school levels. Many of such studies are conducted with a small number of schools which circumvents meaningful comparison at the school level. Designing, conducting, and analyzing CRTSS require accounting to sources of variations related to implementations of the interventions proposed, school’s seasons and years, and the therapists employing the interventions. Other source of variations is related to between cluster (school) variation and within school correlations. Power estimations and sample size calculations for CRTSS require special attention to the nesting design of such trials. We will discuss and present recent studies utilized the CRTSS such as a cluster randomized trial to evaluate external support for the implementation of positive behavioral interventions and supports by school personnel, The Preventing Relational Aggression in Schools Everyday Program and Family-School Intervention for Children with ADHD.
Alfred Inselberg
Tel Aviv University
Israel
Title: Tutorial on Visualization and data mining for high dimensional datasets
Biography:
Alfred Inselberg received a PhD in Mathematics and Physics from the University of Illinois (Champaign-Urbana) and was Research Professor there until 1966. He held research positions at IBM, where he developed a Mathematical Model of Ear (TIME Nov. 74), concurrently having joint appointments at UCLA, USC and later at the Technion and Ben Gurion University. Since 1995, he is Professor at the School of Mathematical Sciences at Tel Aviv University. He was elected Senior Fellow at the San Diego Supercomputing Center in 1996, Distinguished Visiting Professor at Korea University in 2008 and Distinguished Visiting Professor at National University of Singapore in 2011. Alfred invented and developed the multidimensional system of parallel coordinates for which he received numerous awards and patents (on air traffic control, collision-avoidance, computer vision, data mining). The textbook ‘Parallel Coordinates: VISUAL Multidimensional Geometry and its Applications’, Springer (October) 2009, has a full chapter on data mining.
Abstract:
A dataset with M items has 2M subsets anyone of which may be the one fulfilling our objectives. With a good data display and interactivity our fantastic pattern-recognition can not only cut great swaths searching through this combinatorial explosion, but also extract insights from the visual patterns. These are the core reasons for data visualization. With parallel coordinates (abbr. k-cs) the search for relations in multivariate datasets is transformed into a 2-D pattern recognition problem. The foundations are developed interlaced with applications. Guidelines and strategies for knowledge discovery are illustrated on several real datasets (financial, process control, credit-score, intrusion-detection etc.), one with hundreds of variables. A geometric classification algorithm is presented and applied to complex datasets. It has low computational complexity providing the classification rule explicitly and visually. The minimal set of variables required to state the rule (features) is found and ordered by their predictive value. Multivariate relations can be modelled as hyper-surfaces and used for decision support. A model of a (real) country’s economy reveals sensitivies, impact of constraints, trade-offs and economic sectors unknowingly competing for the same resources. An overview of the methodology provides foundational understanding; learning the patterns corresponding to various multivariate relations. These patterns are robust in the presence of errors and that is good news for the applications. We stand at the threshold of breaching the gridlock of multidimensional visualization. The parallel coordinates methodology has been applied to collision avoidance and conflict resolution algorithms for air traffic control (3 USA patents), computer vision (1 USA patent), data mining (1 USA patent), optimization, decision support and elsewhere.
Altaf H Khan
King Abdullah International Medical Research Center
National Guard Health Affairs
Saudi Arabia
Title: Workshop on Choice of agreement in medical imaging: Inter-rater, intra-rater reliability index or receiver operating curve or any other index to rely on
Time : 11:30-13:00
Biography:
Altaf H Khan has completed three Master’s degrees in Biostatistics (2004), Applied Mathematics (1999) and Mechanical Engineering (2003) from the University of Utah. Currently, he is working as a Senior Biostatistician at King Abdullah International Medical Research (National Guard Health Affairs), Riyadh (Saudi Arabia) and prior to that he worked at the University of Utah Hospital and Prince Sultan Cardiac Center. He has many publications in international journals and proceedings.
Abstract:
As advancement in medical imaging has revolutionized the healthcare industry and now burden rests on the shoulders of the biomedical researchers and clinicians to make fruitful inferences from these bulk of the data. The very fundamental question and the crux of the issue is how radiologists as observers could agree on some reliable indices such as intra or inter-rater agreement, Receiver-Operating Curve (ROC) or any other reliable index which would pave the way for making clinical decision in the prognosis of an illness, because FDA (Federal Drug Administration) also strongly requires in clinical trial studies to establish as well as support the efficacy of medical imaging agent. In this paper, an attempt has been made to discuss existing reliability indices such as Cohen kappa, weighted kappa, and kappa used in triage system as well as review has been done for other existing indices. Also, ROC has been discussed with or without any gold standard medical imaging modalities, since there are no unanimous regulations from the manufacturing industries. Using PROC IML, macros has been written to compute reliability index used in triage system which is based upon taking into account the severity of mis-triage in which the reliability index has been calculated by applying an alternative weighting method. Computed reliability indices such as simple kappa, weighted kappa, as well as triage kappa have been compared with those indices that are available in standard statistical software packages namely SAS, R, Stata, etc. Reliability indices based on Bayesian approach have also been discussed.
Al Omari Mohammed Ahmed
Al Baha University
Saudi Arabia
Title: Markov Chain Monte Carlo estimation for Bayesian approach based on type-I censored data
Time : 14:50-15:10
Biography:
Al Omari Mohammed Ahmed has completed his PhD at the age of 31 years from Putra University of Malaysia. He is the head of department of Mathematics in faculty of art and sciences in AlBaha University. He has published more than 12 papers in reputed journals and he interesting in Bayesian statistics and survival analysis study.
Abstract:
This study consider the estimation of Maximum Likelihood Estimator and the Bayesian Estimator using Jeffrey’s prior and Extension of Jeffrey’s prior information of the Weibull distribution with type-I censored data. The shape parameter estimation by maximum likelihood method has been seen that are not available in closed forms, although they can be solved them by numerical methods. Moreover, the Bayesian estimates of the parameters, the survival and the hazard functions we can’t solve it analytical for that Markov Chain Mote Carlo is used, where the full conditional distribution for the scale and shape parameters are obtained via Gibbs sampling and Metropolis-Hastings algorithm following by estimated the survival and hazard functions. The methods are compared to Bayesian using Lindley’s approximation and maximum likelihood counterparts and the comparisons are made with respect to the Mean Square Error (MSE) and absolute bias to determine the best estimating of the parameters, the survival and the hazard functions.
Yanhong Gao
Chinese PLA General Hospital
China
Title: Establishment and verification of miRNAs expression molecules for breast cancer and its distant metastasis
Time : 15:10-15:30
Biography:
Yanhong Gao has completed her PhD from Chinese Academy of Medical Sciences in 2005. She work as an associate chief physician and associate professor at Department of Clinical Biochemistry of Chinese PLA General Hospital since 2005. She is interested in finding tumor new biomarkers for diagnoses and prognostic from blood by advanced technology methods(including Biometrics & Biostatistics). She has published more than 25 papers in journals.
Abstract:
Breast cancer is one of malignant tumors in women. Distant metastasis is the main death cause for breast cancer patients. The early detection of laboratory is the key for prevention and therapy breast cancer. MicroRNAs are a large class of single-strand endogenous non-coding small RNA. It was reported that cells may release tumor endogenous microRNAs into peripheral blood cycle and become circulating microRNAs in tumor genesis and development. We have analysed serum microRNAs of breast cancer and breast cancer metastasis patients by new screening strategy and biostatistics method. We have got hsa-miR-6090 and hsa-miR-451a as candidated molecules of breast cancer and breast cancer metastasis. Therefore, we speculate that hsa-miR-6090, hsa-miR-451a play an important role in the development and metastasis of breast cancer, and further study to verify their biological function will be go on.
Brent Spruill
Walden University
USA
Title: Increasing obesity rates among adolescents in the State of Massachusetts
Biography:
Brent Spruill is an sssistant director player personnel at Colorado Crush. He works with all students in Anatomy and Physiology and Statistics. He is responsible for scouting of all professional football teams all over the country in the National Football League (NFL) and Canadian Football League (CFL) for possible candidates for his organization.
Abstract:
Increasing obesity rates among adolescents in the State of Massachusetts are of concern to public-health professionals. High bullying rates may contribute to obesity. Guided by Maslow's safety component and Bandura's social-cognitive theory, this study investigated a relationship between hours spent television watching, bullying, and meeting physical-activity guidelines among Massachusetts adolescents. The association between the dependent variable (physical inactivity) and the independent variables (hours spent watching television and bullying) was explored using data from the 2009 Massachusetts Youth Risk Behavior Survey. Participants were 2,601 Massachusetts adolescents aged 13 to 18. Statistical analysis included chi-square, the Kruskal-Wallis Test, Mann-Whitney U, and Spearman correlation. Results revealed a significant negative correlation between television watching and physical activity, suggesting that the more hours students spent watching television, the less active they tended to be. The Kruskal-Wallis test showed a significant difference in hours of television watching by level of physical activity. To determine where the statistical differences lay, 3 pair-wise Mann-Whitney U tests were conducted; 2 were shown to be statistically significant. Physical activity and bullying were significantly associated. The results of the Mann-Whitney U test were significant, indicating that levels of activity for students who were not bullied were higher than those for students who were bullied. The social-change potential of this study is a better understanding of the relationship between bullying and physical inactivity among public health professionals in an increased effort to remove barriers to physical inactivity, help limit bullying, and increase health and welfare of adolescents.
Ken Williams
KenAnCo Biostatistics
University of Texas
USA
Title: A workshop on how to do meta-analysis right
Time : 15:45-16:45
Biography:
Ken Williams received a BS in Applied Math from Georgia Tech in 1971 and an MS in Operations Research from the Air Force Institute of Technology in 1980. He served in the US Air Force for 22 years in Computer Systems and Scientific Analysis. He also served 10 years as a Biostatistician at the University of Texas Health Science Center at San Antonio where he remains an Adjunct Faculty Member. He has been a Freelance Biostatistician with KenAnCo Biostatistics since 2007. Designated as a Professional Statistician (PStat) in the inaugural 2011 litter, he has published more than 100 papers in peer-reviewed journals.
Abstract:
This workshop will provide a brief overview of important features of meta-analysis. Topics will include: choosing between a fixed effect and a random effects model; accounting for correlations between statistics being compared; assessing the potential for bias; conducting subgroup analyses; doing meta-regression analysis; comparing the advantages and disadvantages of using published statistics versus individual-level data; doing Bayesian meta-analysis; choosing among available meta-analysis software; and applying parameters estimated by meta-analysis to support public health policy decisions. Examples will be provided from three published meta-analyses. One included all the 12 published reports from epidemiological studies that contained estimates of the relative risks of LDL-C, non-HDL-C, and apoB predicting fatal or nonfatal ischemic cardiovascular events. Another meta-analysis included 7 placebo-controlled statin trials in which LDL-C, non-HDL-C, and apoB values were available. The workshop leader was the lead analyst for these first two sample meta-analyses. The third sample meta-analysis was conducted by the Emerging Risk Factors Collaboration using individual-level epidemiological data from 3 studies which had published the relevant statistics and 23 which had not. All these sample meta-analyses were published in various journals. The workshop will wrap up with a discussion of how irreconcilable conclusions may be derived from different meta-analyses ostensibly pursuing the same objective.
Martial Longla
University of Mississippi
USA
Title: An objective Bayesian estimation of parameters in a log-binomial model
Biography:
Martial Longla received several diplomas at the Peoples' Friendship University of Russia: Teacher of Russian as Foreign Language, Interpreter/Translator with 3 languages (French, Russian, and English) with honors, Bachelor of Sciences and Master of Sciences in Mathematics, and several University and city awards for his leadership in the fight for students rights and promotion of African culture. He completed a PhD program in Moscow on Optimal Control Problems in Infinite Dimensional Spaces in 2008 and moved to the University of Cincinnati where he obtained a PhD in Mathematics in 2013. He joined the Department of Mathematics at the University of Mississippi in August 2013.
Abstract:
The log-binomial model is commonly recommended for modeling prevalence ratio just as logistic regression is used to model log odds-ratio. However, for the log-binomial model, the parameter space turns out to be restricted causing difficulties for the maximum likelihood estimation in terms of convergence of numerical algorithms and calculation of standard errors. The Bayesian approach is a natural choice for modeling log-binomial model, as it involves neither maximization nor large sample approximation. We consider two objective or non-informative priors for the parameters in a log-binomial model: an improper prior, and a proper prior. We give sufficient conditions for the posterior from the improper at prior to be proper, and compare the two priors in terms of the resulting posterior summaries. We use Markov Chain Monte Carlo via slice sampling to simulate from the posterior distributions. An overview of recent contributions to this problem will be provided. We will also present questions involving dependence.
Biography:
Alfred Inselberg received a PhD in Mathematics and Physics from the University of Illinois (UICU). He was Graduate Assistant at the Biological Computer Lab (BCL), where research on Brain Function, Cognition and Learning was carried out (coupled to McCulloch’s Lab at MIT of neural networks fame), and continued at BCL as Research Assistant Professor. During 1966-1995, he was IBM Researcher (reaching a rank just below Fellow) at the Los Angeles Scientific Center and later Yorktown Labs. He developed a Mathematical Model of the (Inner) Ear (TIME, Newsweek 1974, etc.) concurrently teaching at UCLA and USC. He joined the Technion’s faculty 1971-73, Ben Gurion University 1977-83, and is at Tel Aviv University since 1995. He was elected Senior Fellow in Visualization at the San Diego Supercomputing Center (1996), Distinguished Visiting Professor at the Korea University (2008) and National University of Singapore (2011). He invented the multidimensional visualization methodology of Parallel Coordinates which has become widely accepted and applied (air traffic control, data mining, etc.). His textbook is on the subject, published by Springer.
Abstract:
I want to be stunned by a visualization discovery … a WOW moment! And I do not mean that some variable values turned out to be much different than expected or the location of an event was different than expected, etc. But rather that something far-reaching we had no idea existed was found, something like … penicillin! This should be the measure of visualization’s success. And just how do we do that? For one thing luck helps but, as I tell my students, “When you work harder your luck … improves”! For a data-set with M items there are 2M possible subsets anyone of which may turn out to be the one satisfying our objectives. With our fantastic pattern-recognition ability we can cut great swaths through this combinatorial explosion discovering patterns corresponding to relational information from a good data display. This is something that simply cannot be automated … thank goodness! Patterns are geometrical creatures and so we need to learn geometry. Actually from our point-of-view we are not interested in rigid patterns but malleable ones e.g. “gaps” which can be different in shape but are gaps, nonetheless. That is we are really in the topology of the patterns. It has been shown that multidimensional patterns cannot be discovered directly from their points. Rather they can be synthesized from lower dimensional information. Even in 3-D, we learn to look at planes not by their points but by their planar surface/shape consisting of their lines and ditto for surfaces. We need to discuss and adopt a rigorous syllabus for the discipline of Visualization involving geometry, topology and cognition among others. This is our best investment for the future. Research on the geometry and topology induced by ||-coords has made great strides. Many patterns corresponding to multivariate relations have been discovered. We have embarked on a project to transform these results into powerful tools for our exploration and data mining arsenal. They revolutionize the power of modern ||-coords.
Muhammad Salman Bashir
King Fahad Medical City
Saudi Arabia
Title: Biostatistics and clinical research: Psychological burden and low level of knowledge among medical practitioners at King Fahad Medical City, Riyadh
Biography:
Muhammad Salman Bashir has an MSc in Statistics and is a Certified Clinical Research Associate from Canadian Association of Clinical Research. He is working in King Fahad Medical City in the capacity of a Biostatistician Specialist I at the Research Center and has a great seven years of experience of Hospitals and Pharmaceutical products and their local clinical trials. He has more than six publications in local and international journal.
Abstract:
Background: Physicians, particularly those with no formal education in epidemiology and biostatistics, had a poor understanding of common statistical tests and limited ability to interpret study results. Fundamental concept of biostatistics and epidemiology are awful for physicians. If physicians do not understand fully the primary concept of biostatistics and epidemiology, then conclusions reach will be more likely to be wrong. Objective: To evaluate the low level knowledge and awareness of basic and advanced biostatistics and epidemiology among physicians, residents, clinicians and researchers at King Fahad Medical City. Methodology & Design: The cross sectional descriptive study design was used. The survey was completed among 250 participants in this study. Target sample was enumerated of all physicians, clinicians, residents, researchers and interns; both male and female; from different departments who were practicing and worked in their OPD, emergency, clinics and other faculties. Result: The initial pilot survey was completed only 250 participants from 8 departments and 3 faculties. The overall mean percentage corrected answer score based on statistical knowledge and biostatistics of results was 31.8% [95% C.I, 28.6% - 38.2%] in contrast 65.6% [95% C.I, 58.3% - 72.1%] for research fellows and general medicine faculty with research training which is highly statistically significant at (p<0.001). High scores in resident were associated with additional advanced degrees 48.3% [95% C.I, 45.6 – 55.8%] in comparison with 42.5% [95% C.I, 38.3% - 44.6%] at (p<0.001). Conclusion: A large number of medical practitioners had low level knowledge and concept of biostatistics and unable to interpret basic and advanced statistical concept that commonly found in the medical literature. Formalized teaching system of biostatistics and epidemiology will be required during the residency for better understanding and proficient in statistical information.
Edwin M M Ortega
Universidade de São Paulo
Brazil
Title: A power-series beta Weibull regression model for predicting breast carcinoma
Biography:
Edwin M M Ortega achieved a PhD in Statistics from University of São Paulo in 2002 and is a Full Professor at Federal University of São Paulo, Brazil. He published more than 130 papers in internationally referred journals. He has experience in probability and statistics, focusing on parametric inference, acting on the following subjects: distribution theory, survival analysis, residual analysis and sensibility analysis.
Abstract:
The postmastectomy survival rates are often based on previous outcomes of large numbers of women who had a disease, but they do not accurately predict what will happen in any particular patient's case. Pathologic explanatory variables such as disease multi-focality, tumor size, tumor grade, lymphovascular invasion and enhanced lymph node staining are prognostically significant to predict these survival rates. We propose a new cure rate survival regression model for predicting breast carcinoma survival in women who underwent mastectomy. We assume that the unknown number of competing causes that can influence the survival time is given by a power series distribution and that the time of the tumor cells left active after the mastectomy for metastasizing follows the beta Weibull distribution. The new compounding regression model includes, as special cases, several well-known cure rate models discussed in the literature. The model parameters are estimated by maximum likelihood. Further, for different parameter settings, sample sizes and censoring percentages, some simulations are performed. We derive the appropriate matrices for assessing local influences on the parameter estimates under different perturbation schemes and present some ways to assess local influences.
Khalaf S Sultan
King Saud University
Saudi Arabia
Title: Robust regression analysis using re-descending M and MM estimators based on modified Cauchy function
Biography:
Khalaf S Sultan is a Professor at King Saud University, Saudi Arabia. He earned BS in Mathematics from Assuit University Egypt, Master’s degree in Mathematical Statistics from Assuit University, Egypt and PhD in Statistics from Al-Azhar University Egypt under the channel system with McMaster University, Canada. He has published journal and conference papers. His research interests include statistical inference, modeling and simulation, optimization, reliability and mixture models.
Abstract:
The author has proposed the modify Cauchy function that can be used to develop re-descending M and MM estimators in robust regression. The proposed modified Cauchy estimator competes with Tukey’s bi-weight and Qadir’s beta resulting in its enhanced efficiency. In addition, to show the usefulness of the proposed technique, they carry out some Monte Carlo simulation experiments. Further, they apply the findings to some real data set.
Yusuf O B
University of Ibadan
Nigeria
Title: An insight on the misuse of logistic regression model in the face of non-convergence
Biography:
Yusuf O B is a student of University of Ibadan, Nigeria. She has expertise in medical statistics and her research interests include: Mathematical Epidemiology of Infectious Diseases (Malaria), Multilevel Modeling and Analyses of Longitudinal Data.
Abstract:
Background: Logistic regression model is widely used in health research for description and predictive purposes. Unfortunately, most researchers are sometimes not aware that the underlying principles of the techniques have failed when the algorithm for maximum likelihood does not converge. Young researchers particularly postgraduate students may not know why separation problem whether Quasi or Complete occurs, how to identify it and how to fix it. Objective: This study was designed to critically evaluate convergence issues in articles that employed logistic regression analysis published in an African Medical Journal between 2004 and 2013. Methods: Problems of Quasi or Complete separation were described and were illustrated with the National Demographic and Health Survey dataset. Assessment of articles that employed logistic regression was conducted. Results: A total of 581 articles were published, of which 40 (6.9%) used binary logistic regression. However, 24 (60.0%) stated the use of logistic regression in the methodology, while only 3 (12.5%) of these properly described the procedures. None of the articles assessed model fit while majority presented insufficient details of the procedures. In addition, of the 40 that used the logistic regression, the problem of convergence occurred in 6 (15.0%) of the articles. Conclusion: Logistic regression tends to be poorly implemented in studies published between 2004 and 2013. Our findings showed that the procedure may not be well understood by researchers since very few described the process in their reports and may be totally unaware of the problem of convergence or how to deal with it.
- Young Researchers Forum
Session Introduction
Bingjie Li
The University of Texas, USA
Title: Combine genetic, molecular, cellular and statistical analysis to determine pathogen variations
Time : 16:45-16:55
Biography:
Bingjie Li holds BM, MPH degrees and is a PhD candidate working at the University of Texas Health Science Center at Houston under supervision of Dr Zhi-Dong Jiang. She has completed her Clinical Medicine degree from Weifang Medical University and Master of Public Health from the University of Texas School of Public Health. She is currently working on several research projects within Center for Infectious Diseases that explore pathogen variations by a combination of genetic, molecular, cellular and statistical methods.
Abstract:
Individualized target-specific molecular medicine improves the patient managements and effective treatment of diseases in the future. The individualized medicine will emphasize the differences from both host and pathogens for every disease. The diagnosis will be determined at disease molecular levels and the treatments will be applied according the difference of individual patient as well as pathogen. We previously demonstrated a single genetic mutation will alter the disease onset time, severity and location in the host. In this study, we intend to study the variations of a single pathogen. Clostridium difficile (C. difficile) is a gram-positive, anaerobic, spore-forming bacillus that can cause pseudomembranous colitis requiring colectomy and resulting in death in hospitalized patients. Patients receiving the first-line traditional antibiotic treatments for CDAD have initial recurrence at a rate of 25% (RCDAD). Once a recurrence occurs there is a 45% chance of a second recurrence, and after the second recurrence, 65% of patients will have a third recurrence. We hypothesize the C. difficile has different variations and this variation could be the contributor for treatment failure. To test our hypothesis, 148 strains of C. difficile were collected from the hospital. All samples were confirmed the C. difficile presentation by culture and toxin assays. Bacterial DNA was extracted. Their molecular components were further analyzed by PCR amplification using four pairs of primers followed sequencing. Bacteria genetic characteristics were identified according to their molecular finger printing by Multilocus Sequence Typing (MLST) and dendrogram analysis. Although all bacteria are similar to the standard positive control by culture and toxin assays, biostatistical analysis reveals that they belong to different clusters by MLST. The data suggest the bacteria from environmental and infected patients are different. The bacteria in the same cluster still have genetic differences. The bacteria genetic differences may be one of the contributors that cause the treatment failure. The combination of genetic, molecular, cellular and statistical analysis will not only help us to understand the complexity of disease processes but also guide us to develop sufficient personalized disease management as well as discover more effective drugs to treat diseases.
Wang Shao Hsuan
National Taiwan University
Taiwan
Title: Generalized concordance measure: Generalized regression model and dimension reduction
Time : 16:55-17:05
Biography:
Wang Shao Hsuan is currently studying at National Taiwan University, Department of Mathematics in a PhD program. He had published a paper in SCI while pursuing his Master’s degree.
Abstract:
In the scientific research literature, rank-based measures have been widely used to characterize a monotonic association between a univariate response and some transformation of multiple covariates of interest. Instead of using a linear combination of covariates, we introduce a multivariate polynomial score to compute the corresponding concordance index through more general semi-parametric regression models. It involves the estimation for the degree of the multivariate polynomial and the central subspace (CS). To deal with this research issue, we propose a BIC-type estimation approach, which is implemented by an effective computational algorithm, to achieve the model selection consistency.
Yusuf O B
University of Ibadan
Nigeria
Title: An insight on the misuse of logistic regression model in the face of non-convergence
Time : 17:05-17:15
Biography:
Yusuf O B is a student of University of Ibadan, Nigeria. She has expertise in medical statistics and her research interests include: Mathematical Epidemiology of Infectious Diseases (Malaria), Multilevel Modeling and Analyses of Longitudinal Data.
Abstract:
Background: Logistic regression model is widely used in health research for description and predictive purposes. Unfortunately, most researchers are sometimes not aware that the underlying principles of the techniques have failed when the algorithm for maximum likelihood does not converge. Young researchers particularly postgraduate students may not know why separation problem whether Quasi or Complete occurs, how to identify it and how to fix it. Objective: This study was designed to critically evaluate convergence issues in articles that employed logistic regression analysis published in an African Medical Journal between 2004 and 2013. Methods: Problems of Quasi or Complete separation were described and were illustrated with the National Demographic and Health Survey dataset. Assessment of articles that employed logistic regression was conducted. Results: A total of 581 articles were published, of which 40 (6.9%) used binary logistic regression. However, 24 (60.0%) stated the use of logistic regression in the methodology, while only 3 (12.5%) of these properly described the procedures. None of the articles assessed model fit while majority presented insufficient details of the procedures. In addition, of the 40 that used the logistic regression, the problem of convergence occurred in 6 (15.0%) of the articles. Conclusion: Logistic regression tends to be poorly implemented in studies published between 2004 and 2013. Our findings showed that the procedure may not be well understood by researchers since very few described the process in their reports and may be totally unaware of the problem of convergence or how to deal with it.
Anam Riaz
National College of Business Administration and Economics (NCBA&E)
Pakistan
Title: Role of BMS and infrastructure in crude death rate and infant mortality rate
Time : 17:15-17:25
Biography:
Anam Riaz has completed her BS – Statistics at the age of 22 years from GC University Lahore, Pakistan. Currently she is research scholar in National College of Business Administration & Economics (NCBA&E) Lahore. Recently she got the scholarship from Higher Education Commission (HEC) of Pakistan for completing her PhD.
Abstract:
The aim of the study is to investigate the relationship of infrastructure of health sector and basic medical staff with the IMR and CDR, respectively. Another purpose of doing this study is to describe the historical trend of infrastructure in health sector and Basic Medical Staff (BMS). Crude Growth Rate (CGR) and Year on Year (YoY) percentage change of basic medical staff and infrastructure shows that there is downward trend after the period 1995-96. The results of one way ANOVA show that each decade has different growth in infrastructure and basic medical staff. Similarly, regression analysis shows that there is linear relationship among the IMR, CDR, basic medical infrastructure and BMS. The finding of the study indicates that basic infrastructure and basic medical staff is playing an important role in reducing the CDR and IMR of Pakistan.
Serge M A Somda
Centre Muraz
Burkina Faso
Title: An algorithm to evaluate follow-up strategies after primary treatment in oncology by computer simulation
Time : 17:25-17:35
Biography:
Serge M A Somda is graduated in statistics and in public health. He is completing his PhD in Biostatistics at the University of Toulouse. He is also employed as Methodologist in a Health Research Center in Burkina Faso, Centre Muraz, where he is In-Charge of providing a methodological support to the research projects of the Center. He has contributed to several research projects and is author or co-author of some peer reviewed articles.
Abstract:
Organizing the surveillance of patients treated for cancer, for early diagnosis of recurrences, is still a subject of debate. Evidence needs to be highlighted to determine when a particular follow-up strategy is efficient enough to have a significant impact on survival. However, the clinical evaluation of follow-up programs after primary treatment is difficult to undertake. This work proposes an algorithm to evaluate a novel follow-up surveillance strategy after treatment in oncology. A computer based randomized two parallel arms non-inferiority clinical trial is simulated to compare two strategies. Overall survival and cancer specific mortality were the two endpoints evaluated. The methodology of Discrete Events Simulation, based on Patient Oriented Simulation Technique, was used. The natural history of the patient’s disease after primary treatment was generated for each individual. Then, for each scheduled visit date, this history could be modified if a relapse was detected early enough and efficient treatment options are available. An application of the algorithm based on breast cancer data shows its advantages in decision making.