Diagnostic Narrowing¶
Procedure — instantiates Coarse-to-Fine Search
Implements the archetype by starting with broad symptom, signal, or evidence groups and then applying more specific tests to likely diagnostic regions.
Diagnostic Narrowing begins from broad categories of explanation — symptom clusters, failure classes, evidence groups — ranks them by prior likelihood, and then spends specific tests to discriminate within the most likely region. Its distinctive discipline is what it refuses to leave behind: alongside the probability-ranked search runs an explicit rule-out for dangerous or costly low-probability explanations, and an explicit path back to a discarded category when a test result contradicts the working hypothesis. Where a screening funnel is content to let cheap early cuts stand, this procedure treats every early exclusion as provisional, guarded by a can't-miss checklist and a re-opening trigger. It is coarse-to-fine search built for domains where a false negative can be irreversible.
Example¶
A patient arrives in the emergency department with acute abdominal pain. The physician does not order every possible test; she starts coarse, sorting the presentation into broad anatomical and physiological categories — gastrointestinal, genitourinary, vascular, gynecologic — and ranks them by what the history and exam suggest most. The pain's location and the patient's age push "appendicitis" and "ovarian" high; a quick set of labs and a focused exam refine within that region.
But two things run in parallel with the probability search. First, a fixed list of can't-miss diagnoses — a ruptured aneurysm, an ectopic pregnancy, a bowel perforation — gets an explicit rule-out regardless of how unlikely each seems, because the cost of missing one is catastrophic. Second, when a confirmatory test comes back wrong for the leading hypothesis — the appendix looks normal on imaging — she does not push harder on it; she backtracks, re-expands the differential, and re-enters the search at a category she had set aside. The narrowing is fast, but it is reversible, and it is fenced by explicit don't-miss guards.
How it works¶
- Map to coarse categories. Sort the presentation into a small set of broad explanation groups and rank them by base rate and by the specific evidence in hand.
- Refine within the leading region. Choose tests for their power to discriminate among the top candidates, not merely to confirm a favorite — spend detail where it separates hypotheses.
- Run the can't-miss check in parallel. Maintain a fixed list of high-cost explanations that receive an explicit rule-out irrespective of their rank — the false-negative guard.
- Backtrack on contradiction. When refined evidence conflicts with the working hypothesis, re-open the differential and re-enter at a previously deferred category rather than forcing the current one.
Tuning parameters¶
- Initial breadth — how wide the first differential is. Broad protects against tunnel vision but dilutes early effort; narrow is faster but risks locking out the true cause before it is considered.
- Test-selection rule — order tests by discriminating power or by cost/availability. Chasing the most discriminating test converges fastest; chasing the cheapest can waste steps confirming what a decisive test would have settled.
- Can't-miss threshold — how low a probability, at how high a stake, still earns a mandatory rule-out. Lowering it catches more rare catastrophes at the cost of more "unnecessary" tests.
- Re-open sensitivity — how much contradicting evidence forces a backtrack. Trigger-happy re-opening never converges; a sticky one hardens into premature closure.
When it helps, and when it misleads¶
Its strength is that it makes a large explanation space tractable without gambling everything on the first guess: attention concentrates on the likely region while a safety net catches the rare, dangerous exception and a backtrack path recovers from early error. It fits any domain — clinical, incident response, fault diagnosis, investigation — where explanations cluster into categories and missing one is expensive.
Its signature failure is premature closure[1]: the searcher anchors on the first plausible category, reads subsequent evidence as confirmation, and stops — the can't-miss list and the re-open trigger both quietly abandoned under time pressure. The classic misuse is dropping the don't-miss rule-outs to save tests, which is exactly the economy that turns a rare miss into a catastrophic one. The guarding discipline is a structured differential that must be actively closed rather than passively narrowed, forced re-expansion when a confirmatory test fails, and a can't-miss checklist that is procedural, not discretionary.
How it implements the components¶
coarse_representation— the broad symptom/failure/evidence categories are the low-resolution model of the explanation space.refinement_step— discriminating tests applied within the leading region add the targeted detail that separates candidates.false_negative_check— the mandatory can't-miss rule-out is a first-class guard against wrongly excluding a rare, high-cost explanation.backtracking_path— the re-open-on-contradiction rule provides an explicit route back to a deferred category.
It sets no formal scope boundary on the search (search_space_boundary) and enforces no varied-slate requirement (diversity_quota) — both are carried by Portfolio Screening.
Related¶
- Instantiates: Coarse-to-Fine Search — the probability-ranked narrowing that keeps false-negative guards and a backtrack path.
- Sibling mechanisms: Multi-Resolution Search · Coarse Grid Search · Funnel Process · Portfolio Screening · Progressive Candidate Review · Design Downselection · Search Tree Pruning with Refinement
Editorial Notes¶
Form Classification¶
Form family: Protocol, Workflow & Routine
Rationale: Diagnostic Narrowing operates as a repeatable ordered procedure or handoff sequence that coordinates action because it implements the archetype by starting with broad symptom, signal, or evidence groups and then applying more specific tests to likely diagnostic regions.
Independent corroboration: The frozen evidence defines Diagnostic Narrowing as 'Implements the archetype by starting with broad symptom, signal, or evidence groups and then applying more specific tests to likely diagnostic regions', so its operative form is Protocol, Workflow & Routine.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Medicine & Healthcare
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: Clinical diagnosis cohered progression from broad symptom groups through increasingly specific tests while retaining dangerous alternatives until excluded.
Related originating lineages:
- Engineering & Design — Troubleshooting practice supplied coarse-to-fine fault localization outside medicine.
Review resolution: Clinical diagnosis cohered progression from broad symptom groups through increasingly specific tests while retaining dangerous alternatives until excluded. The retained alternate lineages materially shaped the mechanism's form.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
What makes this a distinct sibling is that its false-negative machinery is inseparable from its narrowing: remove the can't-miss check and the backtrack path and it collapses into an ordinary probability sort. The procedure is designed for stakes where the coarse pass must never be treated as a final gate — the archetype's high-stakes caveat is not an add-on here but the point.
References¶
[1] Croskerry, P. "The Importance of Cognitive Errors in Diagnosis and Strategies to Minimize Them". Academic Medicine 78(8), 775–780 (2003). Describes diagnostic premature closure as anchoring early, favoring confirming information, and stopping before a diagnosis is fully verified. registry ↩