JOURNAL OF RESEARCH IN MEDICAL SCIENCES (JRMS)
Year:2003 | Volume:8 | Issue:2|Papers Count:14

Introduction: The map of diseases is usually constructed using the information from diseases incidnce in some regions. Some factors, such as measurment error and rapid variation of diseases rates in different regions make maps so wiggly that their interpretation becomes difficult. Therefore these maps must be smoothed using statisical methods.Methods: Since disease rates of different regions reflect an spatial correlation structure, in this paper the spatial correlation structure of data is specified by fitting a variogram model, then kriging as a best linear unbiased prediction method is used to make a smooth map of diseases. Results: The tuberculosis incidence rates of 262 counties of Iran are used to demonstrate the application and accuracy of the diseases mapping method presented in this paper. The smoothing map of tuberculosis disease, obtained by kriging method shows the geographical trend of the disease in Iran. In this map, central and western regions of Iran have minimum incidence rates, and it gradually increases toward the eastern boundaries.Discussion: The object of this article is introducing kriging method for disease mapping and tuberculosis disease is used to demonstrate the application of this method. There is on dubt that the numerical results of prediction and mapping can be affected by undercount in the smir positive (S+) tuberculosis data, which are gathered by the office for campaigning against diseases. However this method has a wide application in different areas of medical sciences. such as geographical epidemiology of diseases, environmental health and environmental engineering.

In Parallel designs which each patient receives only one treatment we need a large number of patients for assessing treatment effects. Cross-over trials are Known as longitudinal studies.In cross-over trials each patient is his or her own control and hence it is possible to make more precise result, that is, small number of patients is sufficient. In modelling cross-over designs two important issues must be considered: (i) carry-over effect (H) correlated responses for each patient. In this paper, by using generalized linear models (GLM), we present a transition (Markov) model for binary responses which enables us take into account two important issues simultaneously.

The problem of carry-over effect in two period -two sequence cross-over studies has already brought about a lot of discussion and has induced to bring up different views and recommendations. The present study deals with relationship between carry-over effects and sample size in 2x2 cross-over design. It has been shown that when such effects exist, number of samples will differ from the case when there is no carry-over effect. The difference in sampel size has been estimated and has been shown that depends upon the differences in size and direction of carry-over effect and direct treatment effects. The influence of carry-over effects on power of the test also has been considered and its variation has been shown. The results suggest that while determining sampel size for a 2x2 cross-over study, it is matter of wise to take into account any knowledge about carry-over effect difference which is in hand from a previous or pilot study or even from a rational guess, especially if carry-over effects difference and direct treatment effects difference are in the same direction. Theoretical outcomes are used for a practical example and related results have been presented.

One of the most popular models for survival analysis is the Cox proportional hazard model. In this model, the hazard function may depend on unknowon risk factors which are impossible to include them in the model. This will lead to an increase in the variability of responses, which implies biased and misleading estimates of the- parameters of the Cox model. This problem can be overcome by inclusion of a random effect in the Cox model. The modified model is known as frailty model.In this paper, we compare parameter estimates of Cox and frailty models with a simulation study. The results show that in the univariate survival data, as the number of unknown risk factors incrase so do the magnitude of the bias and the mean square error (MSE) of estimators in the Cox model compared to the frailty model. In bivariate survival data, magnitude of the biased and the mean square error of estimator of the Cox model depends on the magnitude of the correlation of survival times. In comparison, the frailty models remove the disadvantage of the Cox model by considering the correlation of the survival times.

Introduction: Meta-analysis techniques are used to estimate an overall effect across a number of similar studies. meta-analysis of clinical trials have been developed rapidly in recent years and several statistical methods are available in this regard. These methods without the primary data and only based on estimate of treatment effect and variances of estimators in different studies combines the results in order to gain more precise estimate of treatment effect.Methods: Statistical model employed for meta-analysis in this research is fixed effect model with homogeneity of variances of treatment effects. six studies examining the effect of aspirin after myocardial infraction have been combined by this method.Results: In all of six studies chosen for meta-analysis comparison of treatment and control group is based on estimates of odds ratios. only in one on studies death rate in treatment group is more than control group but in none of them differences between groups are statistically significant after meta-analysis confidence interval computed for odds ratio is narrower than Cls computed in individual studies which also implies that aspirin effect in reduction of mortality rate is significant.Discussion: Meta-analysis of clinical trials with homogeneity assumption of variances of treatment effect and substituting pooled variance instead of individual variances could leads to gain more precise estimates for treatment effects. it is more efficient when sample sizes in different studies are small, but in this study results were not very different from those gained by previous methods which does not benefits from pooled estimate of variance, because the sample sizes in studies are virtually large.

Introduction: Artificial neural networks mimic brains behavior. They are able to predict and feature recognition and classification. Therefore, neural networks seem to serious rivals for statistical models like regression and discriminant analysis.Methods: We have introduced biological neuron and generalized their function for artificial neurons and described back propagation error algoritm for training of networks in details. Result: Based on two simulated data and one real data we built neural networks by using back propagation and compared them by regression models.Discussion: Neural networks can be considered as a non parametric method for data modeling and seem that they are potentially. more powerful than regression for modeling, but more ambiguous in notation.

Introduction: In survival analysis, determination of sufficient sample size to achieve suitable statistical power is important .In both parametric and non-parametric methods of classic statistics, randomn selection of samples is a basic condition. practically, in most clinical trials and health surveys randomn allocation is impossible. Fixed - effect multiple linear regression analysis covers this need and this feature could be extended to survival regression analysis. This paper is the result of sample size determination in non-randomnized surval analysis with censored and non -censored data.Methods: In non-randomnized survival studies, linear regression with fixed -effect variable could be used. In fact such a regression is conditional expectation of dependent variable, conditioned on independent variable. Likelihood fuction with exponential hazard constructed by considering binary variable for allocation of each subject to one of two comparing groups, stating the variance of coefficient of fixed - effect independent variable by determination coefficient , sample size determination formulas are obtained with both censored and non-cencored data. So estimation of sample size is not based on the relation of a single independent variable but it could be attain the required power for a test adjusted for effect of the other explanatory covariates. Since the asymptotic distribution of the likelihood estimator of parameter is normal, we obtained the variance of the regression coefficient estimator formula then by stating the variance of regression coefficient of fixed-effect variable, by determination coefficient we derived formulas for determination of sample size in both censored and non-censored data.Results: In no-randomnized survival analysis ,to compare hazard rates of two groups without censored data, we obtained an estimation of determination coefficient ,risk ratio and proportion of membership to each group and their variances from likelihood function, when data has censored cases an estimate of the probability of censorship should be considered, after obtaining the varince of maximum likelihood estimator and considering its asymptotic normal distribution and by using coefficient of determination, formulas have been derived. The derived sample size formulas could attain the required power for a test adjuasted for effect of other explanatory covariates.Discussion: application of regression model in non-randomnized survival analysis helps to derive suitable formulas to determin sample size in both randomized and non-randomnized studies in a error level, to attain necessary statistical power. In Coxs semiparametric proportional hazard model ,since the varince of the parameter can not be stated in a simple form ,a simulation model can be used. When the coefficient of determination is partialy large the power bassed on log-rank test overestimates the true value of power, but when coefficient of determination is near to difference between powers decreases zero. By increasing of regression coefficient of determination, the difference between the log-rank test and adjusted coefficient of determination of this paper increases.