C. Ponnuraja ¹, P. Venkatesan ¹

Affiliations
1 Tuberculosis Research Centre (ICMR), Chennai-600 031, IN

Abstract

The demonstration of varying treatment effects among different subjects of patients is an important part of the analysis of clinical trials. But issues of censoring, truncation and inclusion criteria complicate the analysis of clinical trial data. Recent advances in proportional hazard methodologies provide regression diagnostics, improved point and interval estimates of the parameters of survival functions, handling of time dependent covariates in the analysis. This paper discusses the interactions between treatment and patient in the presence of censoring and to account heterogeneity using frailty model. The application of the frailty model with respect to pulmonary tuberculosis data are presented and discussed.

Keywords

Tuberculosis, Cox Proportional Hazard Model, Time-dependent Covariates, Gamma Shared Frailty Model

Full Text

   

P. Venkatesan ¹, R. Srinivasan ¹

Affiliations
1 Tuberculosis Research Centre, ICMR, Chennai – 600 031, IN

Abstract

In this paper, we have used statistical measures and spatial deviational ellipse to determine the spatial pattern of tuberculosis within a Chennai ward population to gain insight into the disease spread. Variogram is used to describe the spatial dependence of tuberculosis in Chennai wards and it is compared with theoretical variogram model of spherical, Gaussian and exponential fitted to tuberculosis data. Arc View GIS 9.2 and SAS software were used for spatial analysis of tuberculosis spread. Data were obtained from District Hospital records for Chennai wards. The results of the spatial pattern revealed that the spread of tuberculosis in Chennai wards have been diverse, with many wards having a low rate of infection and the epidemic being most extreme in slum areas. Variogram increases with distance at small distances and then level off which implies spatial dependence exists between small distance of tuberculosis cases. Spherical model fits data better. Spatial analysis is proved to be more useful for studying spread of tuberculosis analysis and modeling of disease analysis.

Keywords

Bayesian, Disease Mapping, Variogram, Spatial Correlation And Deviational Ellipse

Full Text

   

C. Ponnuraja ¹, P. Venkatesan ¹

Affiliations
1 Tuberculosis Research Centre (ICMR), Chennai-600 031, IN

Abstract

The proportional hazard (PH) model and its extension are used comprehensively to assess the effect of an intervention in the presence of covariates. The assumptions of PH model may not hold where the effect of the intervention is to accelerate the onset of an event. The accelerated failure time (AFT) model is the alternative when the PH assumption does not hold. The aim of this paper is to formulate a model that yields biological plausible and interpretable estimates of the effect of important covariates on survival time. The data consists of 1236 tuberculosis patients admitted in randomized controlled clinical trial. A total of six covariates are considered for modeling. The AFT model gives better prediction than the Cox PH model.

Keywords

Accelerated Failure Time Model, Proportional Hazards Model, Time Dependent Covariate, Tuberculosis

Full Text

   

P. Venkatesan ¹, C. Dharuman ², S. Gunasekaran ³

Affiliations
1 Department of Statistics, Tuberculosis Research Centre, ICMR, Chennai-600 031, IN
2 P. G. Department of Mathematics, Pachaiyappa’s College, Chennai-600 030, IN
3 P. G. Department of Physics, Pachaiyappa’s College, Chennai-600 030

Abstract

In recent years, Fourier Transform Infrared (FT-IR) spectroscopy has had an increasingly important role in the field of pathology and diagnosis of disease states. The principal component regression (PCR) and the partial least squares regression (PLS) are the often proposed methods and widely used in FTIR data analysis, when the number of explanatory variable is relatively large in comparison to the samples as the least squares estimator may fail in such situations. They provide biased estimators with the relatively smaller variation than the variance of the least squares estimators. In this paper, a FTIR diabetes dataset is used in order to examine the performance of the two biased regression models on prediction. The conclusion is that for prediction PCR and PLS provides similar results which require substantial verification for any claims as to the superiority of any of the two biased regression methods.

Keywords

Fourier Transform Infrared, Principal Component Regression, Partial least Square, Diabetes Data

Full Text

   

P. Venkatesan ¹, D. Ramamurthy ², N. Sundaram ³

Affiliations
1 Department of Statistics, National Institute for Research in Tuberculosis, ICMR, Chennai – 600031, IN
2 Department of Mathematics, Sir Theagaraya College, Chennai – 600 021, IN
3 Department of Statistics, Dr. Ambedkar Government Arts College, Chennai - 600 039, IN

Abstract

The current prevalence of HIV infection and the corresponding pattern of incidence from the beginning of the epidemic to the present time are mainly estimated by means of back-calculation method. This back-calculation method reconstructs the past pattern of HIV infection and predicts the future number of AIDS cases with the present infection status. The basic data required for back-calculation methodology is the number of AIDS cases over a period of time. TANSACS publishes the reported number of AIDS cases in Tamil Nadu. In this paper, the various approaches for modeling the incubation distribution are compared using real data under various infection density distributions. The projected minimum and maximum AIDS cases in Tamil Nadu, a southern state of India, based on the reported data are 3702712 and 6936047 respectively. These estimates are based on the unadjusted AIDS incidence data. The purpose of this paper is to review the contribution of back-calculation method to our understanding of the AIDS and to summarize and interpret the epidemiological findings.

Keywords

HIV/AIDS,Incubation Period, Estimation, Infection Distributions, Back Calculation

Full Text

   

P. Venkatesan ¹, N. R. Yamuna ¹

Affiliations
1 Department of Statistics, National Institute for Research in Tuberculosis, (ICMR), Chennai-600 031, IN

Abstract

Decision Trees are a subfield of machine learning technique within the larger field of artificial intelligence. It is a supervised learning technique for classification and prediction. The decision trees are widely used for outcome prediction under various treatments for disease cure, prevention, toxicity and relapse. The aim of the paper is to compare the decision tree algorithms in classifying tuberculosis patient's response under randomized clinical trial condition. Classification of patient's responses to treatment is based on bacteriological and radiological methods. Three decision tree approaches namely C4.5, Classification and regression trees (CART), and Iterative dichotomizer 3 (ID3) methods were used for the classification of response. The result shows that C4.5 decision tree algorithm performs better than CART and ID3 methods.

Keywords

Decision Tree, Data Mining, Cart. C4.5, ID3

Full Text

   

P. Venkatesan ¹, N. Sundaram ²

Affiliations
1 Department of Statistics, Tuberculosis Research Centre, ICMR, Chennai-600 031, Tamilnadu, IN
2 Department of Statistics, Dr.Ambedkar Government Arts College (Autonomous), Vyasarpadi, Chennai-600 039, Tamilnadu, IN

Abstract

The Exponentiated Exponential (EE) model serves as an alternative to Exponential, Weibull and Gamma models. It is observed that EE model has been used in the analysis of complete life time data. In this paper an attempt has been made to study the modeling of censored survival data and the results are compared with other models. Log Likelihood ratio statistic and Cox-Snell residuals are used for the comparisons. The EE model performs better than Exponential and Weibull models. We also fitted Log-logistic model and compared with other models based on Baysian information criterion (BIC) and an information criterion (AIC). The Log-logistic model also performs better than the above models in situations when the censoring is at low level.

Keywords

EE Model, Hazard Function, Life Time Data, Survival Function, Weibull Model

Full Text

   

P. V. Geetha ¹, R. A. Lukshmi ¹, P. Venkatesan ²

Affiliations
1 Department of Mathematics, Meenakshi College for Women, Chennai–600 024, IN
2 National Institutes for Research in Tuberculosis, ICMR, Chennai-600 031, IN

Abstract

This study investigates the application of the hybrid technique Genetic-Neuro approach for Tuberculosis disease classification. Evolutionary algorithms are proved to be the efficient methods for optimization problems and their primary component namely Genetic Algorithm is used to select the significant features for Disease Classification. Artificial Neural Network is used for classification and the training is done by methods like Levenberg Marquardt algorithm. The construction process of the system is illustrated by using tuberculosis disease data. The results reveal that the hybrid technique Genetic-Neural system outperforms the conventional technique Artificial Neural Network for disease classification.

Keywords

Feature Selection, Genetic Algorithm, Neural Network, Tuberculosis

Full Text

   

C. Dharuman ¹, P. Venkatesan ²

Affiliations
1 SRM University, Ramapuram Campus, Chennai - 600089, Tamil Nadu, IN
2 Sri Ramachandra University, Porur, Chennai - 600116, Tamil Nadu, IN

Abstract

Markov Chain Monte Carlo (MCMC) methods have been successfully used to overcome problems involving high dimensional diabetic spectral data. The objective is to show that the onset of a disease changes the relative content of the Bio-Molecules. The problem is formulated and solved in Bayesian frame work by using MCMC algorithm. The efficiency of the models are discussed, evaluated and compared.

Keywords

Bayesian, Diabetic Spectral Data, Markov Chain Monte Carlo, Pattern Analysis

Full Text