Computer-aided detection in screening mammography

entry · p2026-09-06-038 · validated

Mammography sensitivity was 85.3% (95% confidence interval [CI]=83.6–86.9) with and 87.3% (95% CI 84.5–89.7) without computer-aided detection.

tech: cad-mammography
passes the definition as: classifiers (neural nets/statistical pattern recognition) fit to annotated mammogram databases to mark suspicious regions
from: 1998
region: US
domain: health
side: loss
people: 323,973 women (2015)
effect: 2 percentage points (2015) — lower mammography sensitivity when read with CAD (85.3% vs 87.3%), i.e. cancers missed
reversible: yes
who holds the switch: owner:radiology practices and CMS
attribution: named
source kind: primary
the statistic: Across 625,625 digital screening exams in 323,973 US women (2003–2009), sensitivity was 85.3% with CAD and 87.3% without; cancer detection was unchanged at 4.1/1000; the authors concluded CAD 'may result in missed cancers'.
source: https://pmc.ncbi.nlm.nih.gov/articles/PMC4836172/
proposed by: researcher (opus), brief 2026-09-06, era-1990-2011
received: 2026-09-06T09:39:39-04:00
harvest report: impacts/harvest/2026-09-06-era-1990-2011.md

Rulings, in order

2026-09-06T09:39:47-04:00 · validatorpass — quote found on the fetched page (exact match; HTTP 200 text/html; charset=utf-8)
2026-09-06T09:59:36-04:00 · sessionseverity — cancers missed when CAD lowers sensitivity: injury-grade harm with no recourse at the time
2026-09-06T10:02:14-04:00 · sessionseverity — cancers missed; but the count that bore it is the women whose cancer was missed, which the source gives only as a 2-point sensitivity gap — no count, so this cannot band until one exists
2026-09-06T10:02:14-04:00 · sessionseverity — cancers missed; the count that bore it is the women whose cancer was missed — the source gives a 2-point sensitivity gap, not a count — so this cannot band until one exists

The proposal as it arrived (JSON)

{
 "id": "p2026-09-06-038",
 "received": "2026-09-06T09:39:39-04:00",
 "proposer": "researcher (opus), brief 2026-09-06, era-1990-2011",
 "kind": "entry",
 "sha": "01875c0777600849",
 "data": {
  "kind": "entry",
  "tech": "cad-mammography",
  "tech_name": "Computer-aided detection in screening mammography",
  "definition": "classifiers (neural nets/statistical pattern recognition) fit to annotated mammogram databases to mark suspicious regions",
  "from": "1998",
  "region": "US",
  "domain": "health",
  "side": "loss",
  "people": {
   "n": 323973,
   "unit": "women",
   "date": "2015"
  },
  "effect": {
   "n": 2.0,
   "unit": "percentage points",
   "date": "2015",
   "what": "lower mammography sensitivity when read with CAD (85.3% vs 87.3%), i.e. cancers missed"
  },
  "reversible": true,
  "switch": "owner:radiology practices and CMS",
  "statistic": "Across 625,625 digital screening exams in 323,973 US women (2003–2009), sensitivity was 85.3% with CAD and 87.3% without; cancer detection was unchanged at 4.1/1000; the authors concluded CAD 'may result in missed cancers'.",
  "source": "https://pmc.ncbi.nlm.nih.gov/articles/PMC4836172/",
  "quote": "Mammography sensitivity was 85.3% (95% confidence interval [CI]=83.6–86.9) with and 87.3% (95% CI 84.5–89.7) without computer-aided detection.",
  "attribution": "named",
  "source_kind": "primary"
 },
 "from_file": "impacts/harvest/2026-09-06-era-1990-2011.md"
}