University of Melbourne · FACULTY OF COMPUTER SCIENCE

COMP90089 Chap.6 Clinical Task Definition and Study Design

- one subject, every graph, every model, every mark
5 Chapters2-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 6 of 9 · COMP90089

Clinical Task Definition and Study Design

Define index time

The course material gives this chapter a concrete anchor: Week 6 pairs defining the clinical task with study design. That index time anchor controls how target label is explained and how data leakage is tested in changed practice.

Clinical Task Definition and Study Design is a quantitative decision problem built from index time, target label and data leakage.

The aim is to define cohort, comparator, prediction horizon and validation before modelling; a numerical result earns meaning only when the variables, units, assumptions and comparison are all explicit.

Begin with index time: state what quantity it represents, the scale on which it is measured and the condition under which it changes.

Then map every symbol in the Clinical Task Definition and Study Design formula checkpoint to index time before calculation begins.

Next connect target label to the calculation. Show the target label transformation line by line, preserve units and signs, and make any denominator or baseline visible.

A target label calculator output is not a method; the reader must be able to reconstruct why that operation answers the question.

Formula checkpoint: index time

Outcome prevalence
π=NpositiveN\pi=\frac{N_{positive}}{N}

Prevalence is the cohort proportion carrying the defined positive outcome.

Trace target label

Use data leakage to interpret or stress-test the result.

Ask whether the data leakage magnitude is plausible, whether a boundary case behaves as expected and which conclusion would reverse if an assumption changed. This is where computation becomes analysis rather than arithmetic.

When the task is to define cohort, comparator, prediction horizon and validation before modelling, separate inputs supplied by the problem from quantities you derive.

Then report the data leakage result in the language of the course and attach the relevant uncertainty, limitation or decision consequence.

Build a representation check before solving. Put index time, target label and data leakage into a small symbol-and-units table, mark which values are observed and which are calculated, and predict the direction of the result before doing arithmetic.

A sign, scale or unit mismatch in index time then becomes visible at setup instead of being hidden inside a polished final number.

Run one sensitivity test after the baseline answer. Change the input most closely connected to target label, hold the remaining assumptions fixed and recompute only the affected steps. Explain whether the movement in data leakage matches the mechanism.

This target label sensitivity shows which assumption controls the conclusion and prevents a single scenario from being presented as universal.

Test with data leakage

Use a three-column index time error log for COMP90089: translation error, calculation error and interpretation error.

Record the exact line where the target label solution first diverged, rewrite that line, and check it with a limiting case or an independent calculation.

Correcting the first failed target label move is more useful than copying the complete solution again.

A complete response should make the task visible before the detail: identify what must be decided, define the relevant terms, connect the evidence to target label, and use data leakage to test the result.

The final sentence about data leakage should answer the question actually asked rather than merely repeat the topic.

The controlling limit is specific: retrospective convenience can produce a target no clinician can act on prospectively.

Keep that data leakage limit beside the worked example, because it separates a careful COMP90089 answer from one that sounds confident but claims more than the task or evidence supports.

For revision, retrieve index time, target label and data leakage without notes, explain their relationship aloud, then complete a changed version of the application: define cohort, comparator, prediction horizon and validation before modelling.

Record the first failed target label reasoning move and repair it before attempting another case.

In this chapter

What this chapter covers

  • 01

    Index time

  • 02

    Target label

  • 03

    Data leakage

  • 04

    Applying index time

  • 05

    Limits of target label and data leakage

Worked example · free

Estimate prevalence

Q [4 marks]. AskSia-authored practice. A study cohort has 300 outcomes among 5,000 eligible encounters. What baseline must evaluation retain? The step allocation is an independently authored practice structure, not an official marking scheme.
  • 1Divide 300 by 5,000.
  • 1Report 6% prevalence.
  • 1Compare to an always-negative baseline.
  • 1Preserve prevalence in calibration interpretation.
Outcome prevalence is 6%. A model can achieve 94% raw accuracy by predicting no event, so task-appropriate metrics and decision value are required.
Sia tip — Start with the outcome rate before admiring accuracy.
Glossary

Key terms

Index time
Timestamp at which prediction inputs close and the intended decision is made. This chapter uses the concept when students define cohort, comparator, prediction horizon and validation before modelling. Use this definition when the task is to define cohort, comparator, prediction horizon and validation before modelling.
Target label
Operational representation of the future outcome a model is trained to predict. It helps explain the reasoning required to define cohort, comparator, prediction horizon and validation before modelling. Use this definition when the task is to define cohort, comparator, prediction horizon and validation before modelling.
Data leakage
Use of information unavailable at the intended prediction time or contaminated by the target. Its limit matters because retrospective convenience can produce a target no clinician can act on prospectively. Use this definition when the task is to define cohort, comparator, prediction horizon and validation before modelling.
FAQ

Clinical Task Definition and Study Design FAQ

What must be fixed before students define cohort, comparator, prediction horizon and validation before modelling?

Define cohort, comparator, prediction horizon and validation before modelling. Week 6 pairs defining the clinical task with study design. Timestamp at which prediction inputs close and the intended decision is made. This chapter uses the concept when students define cohort, comparator, prediction horizon and validation before modelling.

Can retrospective convenience produce a target no clinician can act on prospectively?

Retrospective convenience can produce a target no clinician can act on prospectively. Operational representation of the future outcome a model is trained to predict. It helps explain the reasoning required to define cohort, comparator, prediction horizon and validation before modelling.

After moving index time earlier, how should a student remove features that become unavailable?

Outcome prevalence is 6%. A model can achieve 94% raw accuracy by predicting no event, so task-appropriate metrics and decision value are required. Retrospective convenience can produce a target no clinician can act on prospectively.

Study strategy

Assessment move

Reconstruct the relationship among index time, target label and data leakage; complete the chapter application without notes; then test the result against this limit: retrospective convenience can produce a target no clinician can act on prospectively.

Working through Clinical Task Definition and Study Design in COMP90089? Sia is AskSia’s AI Computer Science tutor — ask any COMP90089 Clinical Task Definition and Study Design question and get a clear, step-by-step explanation grounded in how COMP90089 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 126 of your University of Melbourne subjects - and 1,000+ Bibles across every Australian university.
Sia - your comp90089 tutor, unlimited, worked the way the exam marks it
The full 2-page Bible + practice bank with worked solutions
Chrome extension - sync your LMS so Sia knows your deadlines
Bilingual EN / Chinese on every Bible and every Sia answer
$0.99 Trial
30-day money-back · cancel in one tap · how it works