Clinical Text AI for Healthcare Leaders

A self-contained, static CPD course for clinical and operational leaders on what natural language processing can really do with hospital records. Twelve sections, each with a quiz, and a printable certificate on completion.

The application is plain HTML, CSS and JavaScript. All course content lives in YAML under content/. Nothing is sent to a server; learner progress is stored in the browser's local storage.

Run locally

Browsers block loading YAML from file:// URLs, so serve the folder over HTTP:

python3 -m http.server 8000
# then open http://localhost:8000/

Deploy to GitHub Pages

1. Push this repository to GitHub.
2. In the repository, open Settings → Pages.
3. Under Build and deployment, choose Deploy from a branch, select the main branch and the / (root) folder, then save.
4. The site will be published at https://<user>.github.io/<repository>/ within a minute or two.

No build step is required. The .nojekyll file tells GitHub Pages to serve the files exactly as they are.

Editing the course

File	Purpose
content/course.yaml	Title, description, learning outcomes, CPD settings (hours, pass mark), parts and the ordered list of sections
content/sections/*.yaml	One file per section: objectives, content blocks and the quiz
index.html, css/styles.css, js/app.js	The player. You should not need to touch these to change content
js/vendor/js-yaml.min.js	YAML parser (MIT licence)

Section file format

id: 04-find                     # must match the id in course.yaml
part: b                         # part id from course.yaml
title: "Find: searching the whole record"
core_question: "Can we find the patients we need to find?"
duration_minutes: 20
objectives:
  - First learning objective
content:
  - type: paragraph             # paragraphs separated by blank lines; **bold**, *italic*, `code` supported
    text: |
      Body text...
  - type: heading               # or subheading
    text: "A heading"
  - type: list                  # style: bullet (default) or number
    items:
      - Plain item
      - term: "Bold lead-in."
        text: "followed by text."
  - type: callout               # kind: key-message | limits | evidence | warning
    kind: key-message
    title: "Optional title"
    text: "Callout text"
  - type: case_study
    title: "Case title"
    text: "Narrative"
    details:
      Capability: "Find"
      Scale: "1,200 patients"
  - type: table
    headers: ["Col A", "Col B"]
    rows:
      - ["a1", "b1"]
quiz:
  intro: "Optional text shown above the questions"
  questions:
    - question: "Question text?"
      type: single              # single (one correct) or multiple (select all that apply)
      options:
        - text: "Correct answer"
          correct: true
        - text: "Distractor"
      explanation: "Shown after submission."

Answer options are shuffled on each attempt. Add shuffle: false to a question to keep its order fixed.

Unlock rules

Parts unlock in order. Within a part, unlock: sequential requires each section to be passed before the next opens; unlock: any lets learners take the sections in any order. The pass mark is set in course.yaml under cpd.pass_mark.

Reviewer access

To open every section without completing the course, visit #/unlock (for example https://<user>.github.io/<repository>/#/unlock). This toggles a preview mode, shown by a banner at the top of the page, and the same address turns it off again. Preview mode never marks a section complete and the certificate still requires every quiz to be passed. The switch is not linked from the interface.

Licence

Course text: © the authors. Player code: MIT. js-yaml: MIT.