Handbook of Statistics Vol 08 - Statistical Methods in Biological and Medical Sciences

The chapters of this volume, divided into four parts, illustrate some recent work
on many of these issues. In the first, five chapters discuss some genetic and
epidemiological problems. Rice et al. reviewed statistical methods relevant for
studying inheritance of qualitative traits. The genetic basis of such traits is often
confounded by age, gender, and environmental covariates, and hence their effects
must be adjusted to tease out the genetic component of variation. These authors
show that the concepts of logistic regression and the LISREL model can be
exploited in such studies. Ewens discusses an important problem of observational
data, where by necessity the investigator must consider only a portion of the
sampling universe. Although the history of ascertainment bias correction is quite
old in the statistical literature, and it led to an important class of distributions,
called Weighted Distributions, Ewens shows that assignment of arbitrary
ascertainment probabilities and non-independence of the probands can easily be
incorporated in complex segregation analysis of qualitative as well as quantitative 
traits. D. C. Rao's chapter deals with the important method of Path Analysis,
which is strictly a causal structural analysis of dependence between variables.
Originally developed by Wright, this methodology had been 're-discovered' and
popularized by Rao, Morton, and others during the mid-seventies. The most
significant genetic project of today is the Human Genome Project, whose ultimate
goal is to understand the genetic organization of the human genome. An important
component of this project is to understand how different genes are linked
(arranged) on chromosomes within a genome. Linkage analysis first started with
experimental organisms where progenies could be generated through directed
breeding experiments, whereby the proximity of genes could be studied by direct
observations on recombinant or non-recombinant progenies. The gene 
organization in humans cannot be studied through such principles for the lack of data
on outcomes of directed matings. As a result, the statistical tools of estimation
and hypothesis testing play a significant role in human linkage analysis. Lathrop
and Lalouel discuss such methods in the fourth chapter. In the last chapter of
Part I, Breslow discusses epidemiological study designs and provides a synopsis
of the current trend of epidemiological research. His chapter also illustrates the
application of categorical data analysis in one of the landmark epidemiological
surveys, the Framingham survey.
The next six chapters in Part II cover applications of statistics in anthropology
and evolutionary biology. Balakrishnan and Ambagaspitiya consider the classical
problem of identification and discrimination with multivariate observations. The
application section of this chapter also illustrates the utility of such methods for
dichotomous traits. Next Chakraborty and Danker-Hopfe discuss the classic
problem of population genetics, the analysis of population structure. These
authors provide a comprehensive review of the existing methods of estimation of
fixation indices, and conclude that in spite of the philosophical differences of
formulation of such estimation procedures, they all yield virtually identical 
estimates. These authors show that a decomposition of diversity indices is a special
form of Categorical Analysis of Variance (CATANOVA). Thompson considers
another important problem, the determination of biological relationships between
individuals from genetic data. The chapter by the co-editors discuss the 
measurement of genetic variation for evolutionary studies, stressing that the different
measures of genetic variation within and between populations can be formulated
under a unified framework. The question of estimation of genetic variation is also
addressed and large sample properties of various measures are described. Saitou's
chapter on phylogenetic tree reconstruction provides another important 
application of statistics in evolutionary studies. The last contribution of this part deals
with some statistical models for sex ratio evolution in which Lessard reviews the
relationships between male and female fitnesses and sex ratio considering genetic
models of evolution of sex ratio.
The next two chapters in Part III relate to the area of cancer biology.
Moolgavkar presents the stochastic models of carcinogenesis that provide a
synthesis of the biology of cell growth, cancer epidemiology, and the stochastic
nature of normall cell differentiation. The application section of this chapter is a
good mixture of experimental and observational epidemiology of cancer risks.
Gart's chapter considers a specialized problem of the application of score 
methodology for confidence interval estimation of one-hit curves that also allows the
comparison of one-hit curves across samples or populations.
Part IV contains three chapters that deal with the subject of medical statistics.
In a case study of -nephropathy, Kardaun's first chapter considers the point
estimation of survival functions. The technique of incorporation of missing value
in the estimation of survival functions brings a distinct novelty in this work. In
the next chapter, Kardaun considers the confidence interval estimation of survival
function using a decision theoretic approach. The concept of effective lifetime
distributions can be of substantial prognostic value in medical statistics. In the
last chapter, Bock and Toutenburg discuss the important subject of sample size
determination in clinical research, examining the sampling adequacies and 
limitations in medical statistical studies.
In summary, the wide variety of applications of statistical methodologies is
conspicuously depicted through the sixteen chapters. Most of the chapters discuss
the statistical principles in conjunction with specific applications. As a result
biological as well as statistical readers should benefit from this exposition, and
obtain the current state of knowledge in these areas. Many authors emphasized
the open problems that might lead to further work on these subjects.