Fundamentals of Medical Statistics

Module code: MD7440
Module coordinator: Stephanie Hubbard

This module provides an overview of the basic statistical methods and approaches to inference applied to medical/health data and the use of statistical software packages.

Statistical Computing

You will be introduced to basic data handling skills using the command languages of Stata and R and will have the opportunity to apply these skills.

Topics covered

  • Writing programs in Stata and R for:
  • Data entry
  • Data manipulation
  • Data analysis
  • Graphs
  • Automation of routine tasks

Statistical Methods

In this teaching week we will review simple traditional statistical methods, apply them to medical/health data sets and discuss the implications of the findings.

Topics covered

  • A review of probability and distribution theory and how probability is communicated in medicine.
  • The importance of sampling variation, how it can be quantified and its role in confidence intervals and hypothesis testing. 
  • Application of a number of simple exploratory and statistical methods to medical/health data using statistical software and reporting on the findings. These methods include:
    • methods to compare group means, variances and proportions
    • correlation and simple linear regression
    • non-parametric methods
    • measuring agreement
    • diagnostic tests.

Statistical Inference

This teaching week introduces the different approaches to statistical inference and contrasts the traditional approach to statistics based on repeated sampling with the likelihood and Bayesian approaches. The theoretical application of those ideas requires a basic knowledge of algebra and calculus. Programming is also an important aspect.

Topics covered

  • Likelihood, maximum likelihood estimate (MLE) and the information
  • Log likelihood curves and support intervals
  • MLE and the observed information for simple distributions
  • Likelihood ratio tests, Score tests and Wald tests and the connection between them
  • Bayes' theorem for both binary and general quantities
  • The nature and source of prior distributions & how to interpret posterior distributions
  • Conjugate models
  • Comparison and contrast of the different approaches to statistical inference

Learning

  • 25 one-hour lectures
  • 30 one-hour practical sessions

Assessment

  • Exam, 90 minutes (50%)
  • Coursework 1 (Statistical Computing) (10%)
  • Coursework 2 (Statistical Methods) (20%)
  • Coursework 3 (Statistical Inference) (20%)