Computer-aided polyp detection in colonoscopy (CADe)

entry · p2026-09-06-292 · validated

Twenty-eight RCTs were selected for inclusion involving 23,861 participants.

tech: learned-colonoscopy-cade
passes the definition as: a convolutional detector fit to labelled endoscopy video that marks suspected polyps on the live image while the endoscopist withdraws the scope
from: 2019
region: world
domain: health
side: benefit
people: 23,861 participants (2024-02)
effect: 20 percent (2024-02) — higher adenoma detection rate with CADe than with unassisted colonoscopy, pooled across 28 randomised controlled trials (risk ratio 1.20, 95% CI 1.14-1.27), with a 55 percent lower adenoma miss rate
reversible: yes
who holds the switch: nobody
attribution: named
source kind: primary
the statistic: Systematic review and meta-analysis in Gastrointestinal Endoscopy (2025), searching to 15 February 2024: 28 randomised controlled trials, 23,861 participants, a 20 percent increase in adenoma detection rate and a 55 percent decrease in adenoma miss rate, the effect holding for expert endoscopists (19 percent). A GI Genius-only meta-analysis of seven trials (n=9,639) finds the same direction (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12616575/). Adenoma detection rate is the quality measure most tightly linked to later colorectal cancer and death from it, which is why this is not a workflow statistic. In severity, most of the 23,861 got an ordinary benefit (a better-quality examination); for the minority whose adenoma was found and removed rather than missed, the effect is a cancer not had, which I would call severe or absorbing. The switch is 'nobody' because this is many vendors in many hospitals, with no single operator who can turn it off. There is no single operator count of people scoped with CADe worldwide; this is the trial population, which is a floor.
source: https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:39216648&format=json&resultType=core
proposed by: researcher (opus), brief 2026-09-06, severe-benefits
received: 2026-09-06T17:40:15-04:00
harvest report: impacts/harvest/2026-09-06-severe-benefits.md

Rulings, in order

2026-09-06T17:40:16-04:00 · validatorpass — quote found on the fetched page (exact match; HTTP 200 application/json;charset=UTF-8)
2026-09-06T17:40:46-04:00 · sessionseverity — A better-quality colonoscopy for the 23,861 trial participants: for most an ordinary benefit; for the minority whose adenoma was removed rather than missed it is severe, but no source counts them. Ordinary on the trial population, switch nobody (many vendors, no single operator), 2019–24.

The proposal as it arrived (JSON)

{
 "id": "p2026-09-06-292",
 "received": "2026-09-06T17:40:15-04:00",
 "proposer": "researcher (opus), brief 2026-09-06, severe-benefits",
 "kind": "entry",
 "sha": "33b63add8222f4cc",
 "data": {
  "kind": "entry",
  "tech": "learned-colonoscopy-cade",
  "tech_name": "Computer-aided polyp detection in colonoscopy (CADe)",
  "definition": "a convolutional detector fit to labelled endoscopy video that marks suspected polyps on the live image while the endoscopist withdraws the scope",
  "region": "world",
  "domain": "health",
  "side": "benefit",
  "from": "2019",
  "people": {
   "n": 23861,
   "unit": "participants",
   "date": "2024-02"
  },
  "effect": {
   "n": 20,
   "unit": "percent",
   "date": "2024-02",
   "what": "higher adenoma detection rate with CADe than with unassisted colonoscopy, pooled across 28 randomised controlled trials (risk ratio 1.20, 95% CI 1.14-1.27), with a 55 percent lower adenoma miss rate"
  },
  "reversible": true,
  "switch": "nobody",
  "statistic": "Systematic review and meta-analysis in Gastrointestinal Endoscopy (2025), searching to 15 February 2024: 28 randomised controlled trials, 23,861 participants, a 20 percent increase in adenoma detection rate and a 55 percent decrease in adenoma miss rate, the effect holding for expert endoscopists (19 percent). A GI Genius-only meta-analysis of seven trials (n=9,639) finds the same direction (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12616575/). Adenoma detection rate is the quality measure most tightly linked to later colorectal cancer and death from it, which is why this is not a workflow statistic. In severity, most of the 23,861 got an ordinary benefit (a better-quality examination); for the minority whose adenoma was found and removed rather than missed, the effect is a cancer not had, which I would call severe or absorbing. The switch is 'nobody' because this is many vendors in many hospitals, with no single operator who can turn it off. There is no single operator count of people scoped with CADe worldwide; this is the trial population, which is a floor.",
  "source": "https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:39216648&format=json&resultType=core",
  "quote": "Twenty-eight RCTs were selected for inclusion involving 23,861 participants.",
  "attribution": "named",
  "source_kind": "primary"
 },
 "from_file": "impacts/harvest/2026-09-06-severe-benefits.md"
}