Biostatistics for health studies: the course in three chapters

A test with a sensitivity and a specificity of 99% can be wrong about ten times out of eleven when it is positive, if the disease is rare: this course gives the statistics a future clinician needs to read a study, interpret a test and decide with data. Probability, diagnostic tests, descriptive statistics, the normal distribution, hypothesis tests, correlation, regression and χ².

Start: chapter 1 →

What the course contains

First-year health studies (PASS, UE8) · 3 chapters, 246 slides in French · summaries, experiments and FAQ in English · CC BY 4.0

Goal
Read a study, interpret a diagnostic test and choose a simple statistical test.
Prerequisites
High-school mathematics: fractions, powers, reading graphs.
Suggested path
Chapter 1 (probability, Bayes, ROC curve with the interactive tool), chapter 2 (normal distribution with the interactive table, tests), chapter 3 (correlation, regression, chi-squared). Each chapter: the 10-point summary, the slides, then the FAQ.
  • Why statistics in health: variability, therapeutic adjustment, correlation and causation, hidden variables.
  • Probabilities: random experiments, events, counting, conditional probabilities, the Bayesian approach.
  • Diagnostic tests: contingency tables, sensitivity, specificity, predictive values, ROC curves.
  • Descriptive statistics: measurement scales, graphs, central tendency, dispersion, skewness.
  • The normal distribution: standardization, the z table, intervals.
  • Hypothesis tests: null and alternative hypotheses, risks, comparison of means.
  • Pearson correlation, linear regression, coefficient of determination and significance.
  • χ² goodness-of-fit and independence tests, with exercises.

Try it: ROC curve · Normal distribution · Student t-test · Influential point in regression · Significance and R² · χ² test · Regression intervals

The words of the course
PASS, UE8
The French first year of health studies; UE8 is its teaching unit 8.
Sensitivity, specificity
The share of sick people whose test is positive; the share of healthy people whose test is negative.
Positive predictive value
The probability of being sick when the test is positive. It depends on prevalence, the share of sick people.
ROC curve
Sensitivity against 1 − specificity as the test threshold varies.
p-value
The probability, if the null hypothesis is true, of a result at least as extreme as the one observed.
Hidden variable
An unmeasured factor that can create an apparent link between two variables.
χ²
A statistic comparing observed counts with expected counts.

Part of the course comes from the work of Dominique Barbolosi, mathematician, Aix-Marseille Université.

Archived and citable on Zenodo, DOI 10.5281/zenodo.23620236

In brief

  • A biostatistics course in three chapters and 246 slides, for the first year of health studies (PASS, teaching unit UE8). The chapters are read here in English; the slides are in French.
  • It goes from probability and Bayesian reasoning to diagnostic tests (sensitivity, specificity, ROC curves), descriptive statistics, the normal distribution and hypothesis tests, then correlation, linear regression and χ² tests, with medical examples.
  • Course by Cyril Voyant, trained as a medical physicist, under CC BY 4.0: each chapter has a ten-point summary with the formulas, experiments to manipulate, a FAQ, and a link to its slides and PDF in French.

These slides are not a handout: they accompany the lectures and cannot replace them (disclaimer of the course). Teaching material, not clinical advice. The slides and the PDF are in French. License CC BY 4.0: free reuse, including modification and commercial use, with credit to the author (Cyril Voyant).

Going further