Determination of sample size to achieve a predefined precision or power
The course will provide an overview on the methods to compute sample size necessary to achieve an adequate precision in estimating proportions or an adequate statistical power to compare the means of two independent or dependent samples, or the proportions of two independent samples. The concepts of statistical power, statistical precision, sample size and effect size will be reviewed. Statistical power computations will be illustrated by Excel files and Stata software. Participants will be encouraged to bring power calculation problems, to be used as examples in the class.
Generalized Linear models: logistic regression, loglinear model, Poisson model
The course will cover a very important family of models in statistics, the so-called, Generalized Linear Models (GLM). We will study the theoretical aspects of different extensions to the linear regression model framework.
Some concrete goals are:
To extend the linear regression model to models that can handle a binary, count and a non-Gaussian response. Course topics include: logistic regression, Poisson regression, log-linear models, contingency tables.
To deal with over-dispersion in the data.
The course aims at providing the theoretical and practical tools for the analysis of survival data in human populations and the associated risk factors, i.e. skills in the field of epidemiology, biostatistics and computer science applied to the analysis of biomedical data (programming syntax of a statistical software).
The course is structured in theoretical and practical lessons (12h) on the use of a statistical software (STATA) for the quantitative analysis of biomedical data. The teaching material is made available to the students on the e-learning web page of the course (Moodle platform).
General concepts of survival analysis: endpoint, survival time, survival function, hazard function.
Univariable analysis: Kaplan-Meier estimator, median survival time, Mantel-Haenszel test.
Multivariable analysis: Cox regression model.
Study design in observational and experimental research
The course will provide an overview on the main designs applied in observational (cross-sectional, cohort, case-control, ecologic) and experimental (parallel groups, cross-over, factorial) research. Strengths and limitations of these designs will be discussed.
Introduction to Meta-analysis, Focused on Medical Research (literature review, data collection, database construction)
The following topics will be covered: formulation of the review question, searching of literature, quality assessment of studies, data extraction, flow-chart of article selection process, according to PRISMA rules.
Application of meta-analysis to the epidemiological or medical field
The course will present examples of meta-analyses performed in the medical field, with special reference to: publication/small series bias; prevalence estimates; evaluation of diagnostic tools.