Lesson 125 of 142 in this sequence

88%

Not completed yet

Saved on this device only — not synced to an account yet.

Learning path

Step 125 of 142

Continue the sequence

ResumeReturn to this step ReviewRevisit the core idea RemediateGet repair practice Next stepChoose the next lesson manually

Completion

Review objectives and instruction
Complete the activity with manual learner confirmation
Answer every understanding-check item
Show completion feedback and reward only after the learner finishes the check

Progress

Step 125/142
15 content types
AI help source context available
Auto-advance blocked

Review objectives and instruction

AI help

Available only with source grounding, privacy limits, age controls, and measurement.

AI source context

Ruling In and Ruling Out: How Diagnostic Tests Perform · Diagnostic Reasoning · Sensitivity · Specificity

Evidence: local lesson AI-help metadata only; source-grounding, privacy, age-safety, and measurement hooks do not prove real model quality, production telemetry, child AI approval, or AI tutor production readiness

Production AI readiness not claimed

AI source context

Assessment shape

Local lesson assessment shape is complete.

Local shape only; production answer-attempt telemetry, live ownership/RLS evidence, mastery interpretation, release, and production readiness are not claimed here.

Objectives: 5
Activity: Prompt ready
Quiz items: 3 items, 3 answer keys, 3 explanations
Reward: Completion reward ready

Assessment progress stays manual: do not auto-advance through uncertain answers, failed writes, consent, media, payment, learner choice, or AI-assisted checks.

Learning-loop evidence: Content-map contract only; persistence and production learner-progress writes require route/API evidence.

Manual policy: assessment, media, consent, payment, learner choice, failed writes, and AI-uncertain help stay under learner control.

Prerequisite: Complete or review Respiratory Therapy and Mechanical Ventilation

AI guardrails

Privacy

Do not send raw child free text without consent
Use source excerpts and non-identifying progress state only
Asset is not registry-gated

Age safety

Apply academy audience controls before AI-help claims
Keep child learner AI paths behind guardian/product review evidence before readiness claims

Measurement

Log AI help usage by surface and asset slug
Evaluate answer quality against source-grounded rubric before production-readiness claims
Tie remediation prompts to completion, retry, or review outcomes

Local AI-help guardrails only; they do not prove guardian approval, production telemetry, model evaluation, release, deployment, or AI tutor production readiness.

Source and rights

Source path: koydo-app/scripts/curriculum-authoring
License: ai-generated-koydo
Attribution: Attribution decision needs owner review
AI provenance: claude-opus-4-8

Source review gates: attribution_missing

Local source evidence only; this does not prove publication rights, attribution approval, AI provenance clearance, release, deployment, or production readiness.

Route review

Current route: /library/academy/medicine_clinical/en/lesson/sensitivity-and-specificity
Route source: Fallback route needs owner canonical review
Candidate route: /library/academy/medicine_clinical/en/lesson/sensitivity-and-specificity

Local route evidence only; fallback candidates need owner-approved canonical route policy before route writes, production discoverability, or readiness claims.

Content gates: canonical_target_web_missing_uses_library_fallback

lessonslessonslesson objectivesmedia scenesdrillspractice setsquizzesfeedbackcharacterscitationsconcept mapsprogressresumereviewremediationnext steps

Ruling In and Ruling Out: How Diagnostic Tests Perform

with Medi

Medi stands at a hospital workstation surrounded by printed lab reports, pointing to a two-by-two grid drawn on a whiteboard while a triage nurse looks on, morning light streaming through a clinic window.

Explain what sensitivity measures and why a high-sensitivity test is preferred for screening serious diseases.
Explain what specificity measures and why a high-specificity test is preferred for confirming a diagnosis.
Identify true positives, false positives, true negatives, and false negatives in a two-by-two contingency table.
Compare how lowering a test threshold increases sensitivity while decreasing specificity.
Calculate sensitivity and specificity from a completed two-by-two table.

Key terms

Sensitivity: Proportion of truly diseased people the test correctly flags positive
Specificity: Proportion of truly healthy people the test correctly clears negative
False negative: A diseased patient the test wrongly labels negative, a missed case
Test threshold: The cut-off value above which a result is called positive

The Sensitivity-Specificity Trade-Off

Sensitivity and specificity are intrinsic test properties tied to where the positive threshold is set, and they pull against each other. Lowering the cut-off catches more true cases and raises sensitivity, but it also sweeps in more healthy people, lowering specificity. Raising the cut-off does the reverse. Because no single threshold maximizes both, clinicians choose the operating point that fits the clinical stakes: maximize sensitivity when a missed case is catastrophic, maximize specificity when a false alarm triggers harmful treatment.

Choosing a Test for the Task

The mnemonics SnNout and SpPin encode the practical rule. A highly sensitive test that returns negative rules a disease out, making such tests ideal for screening serious or contagious conditions where misses are unacceptable, like HIV or a cancer panel. A highly specific test that returns positive rules a disease in, making such tests ideal for confirmation before committing to a major or risky intervention. Real diagnostic strategies often pair a sensitive screen with a specific confirmatory test to capture the strengths of both.

Worked examples

Calculate sensitivity and specificity from a table

Read the two-by-two cells: 80 true positives, 10 false negatives, 100 true negatives, 10 false positives, totaling 200 patients.
Apply sensitivity = TP divided by (TP + FN) = 80 / (80 + 10) = 80/90.
Compute specificity = TN divided by (TN + FP) = 100 / (100 + 10) = 100/110.
Convert to proportions: sensitivity is about 0.89 and specificity is about 0.91.

Answer: Sensitivity is approximately 89% and specificity is approximately 91%.

Hey, I'm Medi — let's talk about one of the most practical ideas in all of medicine: how do we know if a test is actually good? Every diagnostic test divides patients into two groups: those it calls positive (disease likely present) and those it calls negative (disease likely absent). Reality also divides patients: they either truly have the disease or they do not. When you cross those two splits, you get four cells. **True Positive (TP):** Test says positive. Patient actually has the disease. Great — we caught it. **False Negative (FN):** Test says negative. Patient actually has the disease. We missed it. Dangerous. **True Negative (TN):** Test says negative. Patient truly does not have the disease. Also great — correctly cleared. **False Positive (FP):** Test says positive. Patient does not have the disease. We alarmed them unnecessarily. **Sensitivity = TP ÷ (TP + FN)** This is the fraction of ALL truly sick patients the test correctly flags. A test with 95% sensitivity misses only 5% of real cases. High sensitivity = low miss rate. **Specificity = TN ÷ (TN + FP)** This is the fraction of ALL truly healthy people the test correctly clears. A test with 98% specificity falsely alarms only 2% of healthy people. High specificity = low false alarm rate. Here is a pair of memory hooks to lock these in: **SnNout** — a highly **S**e**n**sitive test, when **N**egative, rules **out** disease. **SpPin** — a highly **Sp**ecific test, when **P**os**i**tive, rules **in** disease. These mnemonics are used in clinical reasoning every day. Now here is the critical trade-off: if you lower the threshold that makes a test call 'positive' (say, lower a blood glucose cut-off), MORE sick people get caught — sensitivity rises. But now some healthy people also cross that threshold — specificity falls. The two measures pull in opposite directions. You cannot freely maximize both at once. Practical rule of thumb: use a high-sensitivity test when you absolutely cannot afford to miss a case — think HIV screening or a cancer panel. Use a high-specificity test to confirm a diagnosis before committing to a major treatment.

Activity

A new strep throat rapid test was run on 200 patients; sort each patient result into the correct cell of the two-by-two table.

Practice

Given 45 true positives and 5 false negatives, calculate the sensitivity of the test.

Explain why lowering a test threshold raises sensitivity but lowers specificity.

Common mistakes to avoid

A highly sensitive test confirms disease when positiveSensitivity governs ruling out on a negative; confirming in requires high specificity instead.
Sensitivity and specificity can both be maximized freelyThey trade off through the chosen threshold, so improving one usually degrades the other.

Check your understanding

A screening test for tuberculosis has sensitivity 0.97 and specificity 0.70. A patient tests negative. Which conclusion is best supported?

A laboratory lowers the blood-glucose cut-off for a diabetes screening test from 126 mg/dL to 100 mg/dL. What is the most likely effect on test performance?

In a study, 50 patients with confirmed disease tested positive and 5 with confirmed disease tested negative. What is the sensitivity of this test?

Recap

Sensitivity measures detection of true cases and rules disease out when negative, while specificity measures correct clearance and rules disease in when positive; the two trade off through the threshold, so test choice follows the clinical cost of misses versus false alarms.

Reflect

Where in everyday life have you faced a sensitivity-versus-specificity trade-off?