#courses

OPEN CURRICULUM · R · QUARTO · MIT

#courses

A four-course ladder from first principles to modern statistical learning, built for biomedical researchers who want to stop guessing about statistics.

What this is

This site hosts a complete, opinionated biostatistics programme written in R and Quarto. It runs from the scientific process and data hygiene through regression, study design, causal inference, and modern statistical learning. Every lab is runnable on a laptop, every concept is introduced by way of a plot before a formula, and every inference lab follows the same five-step template so the shape of statistical reasoning becomes familiar long before the mathematics stops being intimidating. The curriculum is MIT-licensed and self-contained: no external data files, no institutional login, no exams. Clone the repository, open an R session, and work through it.

Who this is for

The primary audience is PhD students, clinicians, postdocs, and research engineers in the biomedical sciences — anyone who needs to read a methods section critically, design a defensible study, analyse their own data, and report the result without sweeping the uncertainty under the rug. A working knowledge of basic R and high-school algebra is enough to start Course 1. The later courses assume familiarity with regression but build everything else from the ground up.

The four courses

Course 1 · Introductory

Foundations of Biostatistics with R

The scientific process, data hygiene, probability, sampling, and the core hypothesis tests that anchor everything else. 4 weeks · 20 labs.

Course 2 · Intermediate

Regression, ANOVA & Model Diagnostics

Linear models, ANOVA, GLMs, diagnostics, calibration, and honest model evaluation. 4 weeks · 20 labs.

Course 3 · Advanced

Study Design, Longitudinal Data & Causal Inference

Designing studies; handling missing, clustered, and time-to-event data; and making causal claims with care. 4 weeks · 20 labs.

Course 4 · Specialist

Modern Statistical Learning & High-Dimensional Biomedicine

Regularisation, tree ensembles, Bayesian modelling, omics, and reproducibility at scale. 4 weeks · 20 labs.

The five-step template

Every inference lab marches through the same five steps: Hypothesis → Visualise → Assumptions → Conduct → Conclude. Workflow labs use the variant Goal → Approach → Execution → Check → Report. The sequence is not arbitrary: it mirrors the order in which a Methods section is written. State what you are asking, look at the data before you model it, check whether your chosen method is defensible, run it, and then write down what you found with an effect size and an interval. After twenty repetitions of this rhythm the scaffolding becomes invisible and the statistics starts to feel like a thought process rather than a recipe.

One source, two formats

Every lab in this curriculum is written once and rendered twice. A single .qmd file produces a long-form article for reading and a Reveal.js slide deck for teaching. The article carries the prose, the code, and the discussion; the deck distils the essentials into slides you can present at a group meeting or a seminar. One quarto render builds both.

Further along

• Get Started — install, clone, and render the site locally.
• Schedule — a linked index of every lab, every deck, every script.
• Cheatsheets — sixteen dense one-pagers, one per week, HTML + PDF.
• Interactive apps — Shiny companions to the labs.
• Research workflow, Decision tree, Glossary, Common errors, Writing a report.
• References — the sources and teaching ecosystems we drew on.
• Acknowledgements — the full list of people whose work made this curriculum possible.