W8 · Model colloquium, decision and review

Workshop W8, teaching units 29–32: six project defences with critical perspectives, a comparison of the consequences of wrong decisions, the individual examination, peer revision and anonymous evaluation.
Author

R. Heller

Part IV · Translating models into practice · Level: independent application · Teaching units: 29–32 (4 × 45 min) · Required reading: core readings of B27 and B28 (about 20 minutes) · Project material: colloquium, feedback and evaluation

Guiding question

Which recommendation can we defend today, under which conditions, against which counter-position, and what will we check next?

Learning objectives

By the end of this workshop you can

  1. defend a model and a recommendation in ten minutes and answer critical questions in four;
  2. take a critical perspective (clinic, patients, regulation, data protection, security, methods) on another group’s project and formulate a concrete change request;
  3. compare the consequences of a wrong decision across clinic, laboratory and care system and state an override or stop condition;
  4. revise your recommendation after peer review and document what changed and why;
  5. transfer the approach to a new research or care problem of your own choice.

Preparation

Teaching units

Unit Task Minutes Observable evidence
29 Groups 1–3 each defend their model for ten minutes and answer questions for four minutes. The other groups each represent one critical perspective. 45 Three assessable presentations and one recorded change request each.
30 Groups 4–6 present in the same format. Everyone compares the consequences of a wrong decision across clinic, laboratory and care system. 45 A criteria-based comparison and a reasoned override or stop condition.
31 Individual examination with 20 case-based single-best-answer questions in 25 minutes, including briefing, distribution and collection. 45 Individual examination performance. The public course offers specially developed practice cases instead (self-test).
32 A peer checks each revised recommendation and its evidence. Anonymous evaluation, review of W1 and choice of a next research or care problem. 45 A revision note and an individual transfer task.

Time plan for units 29 and 30

Three groups fit into 45 minutes: 3 × (10 minutes presentation + 4 minutes questions) = 42 minutes, plus 3 minutes for changing over. The course team keeps time strictly; the questions are asked by the groups holding the critical perspectives, in an order drawn at the start.

About the examination (unit 31)

The examination covers all four parts with the distribution shown in the syllabus. Questions are new cases; none of them is published. The formative self-test shows the type of reasoning expected, with explained answers. Assessment and participation conditions for a particular run of the course are announced by the course team before the course starts.

Review and short case B (unit 32)

Rate the four statements from the entry self-assessment in W1 again, anonymously. Then work on short case B for five minutes, on paper and without help:

A clinical team reports “0.91 AUC on 3,000 video frames” for a model that detects an abnormal gait pattern in videos from 15 patients. Frames from every patient were used both for training and for testing. A week later a colleague notices that a screenshot from one of the videos, showing a patient’s face and the ward’s name board, was posted in a group chat that includes an external collaborator.

  1. What is the independent number of units in the test? (b) What does the 0.91 measure? (c) Name the first two steps after the screenshot is discovered.

The course team compares the anonymous answers to short cases A and B. The comparison is used to revise the course; it is not a measure of individual performance, and a difference between before and after is not proof that the course caused it.

Materials

Four perspectives in this workshop

The critical perspectives in units 29 and 30 are assigned so that every project is questioned on ethics, regulation, data protection and security, plus clinical use and methods. Each question is recorded with the answer and, where needed, a change request.

After the course

The transfer task from unit 32 is yours. The case atlas in B28 helps you find the method that fits a new problem, starting from the decision rather than from the data type.

Further reading