Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1829
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-18T18:12:27.981167Z
status: published
lens: g_space
sector: health
headline: Automated Calculation of the Pediatric Emergency Care Applied Research Network (PECARN) Intra-Abdominal Injury Rule from Unstructured Emergency Department Notes Using a Large Language Model
dek: Objective: The Pediatric Emergency Care Applied Research Network (PECARN) intra-abdominal injury clinical prediction rule identifies children at very low risk for clinically significant intra-abdominal injury after blunt abdominal trauma. However, retrospective evaluation of the rule's use in practice requires resource-intensive chart review. We investigated whether large language models (LLMs) can automatically calculate PECARN intra-abdominal injury risk scores from real-world unstructured emergency department…
gain_title: An LLM read unstructured emergency notes and produced PECARN intra-abdominal injury risk scores closely matching clinician-derived scores, which could reduce the chart review needed to evaluate how the rule is used.
problem_title: (none)
trace_subject: (none)
gain_reading: An LLM read unstructured emergency notes and produced PECARN intra-abdominal injury risk scores closely matching clinician-derived scores, which could reduce the chart review needed to evaluate how the rule is used.
gain_evidence: the LLM calculated binary PECARN risk scores with 95.4% accuracy | Agreement by weighted Cohen's kappa between multicategorical LLM-derived risk scores and clinician-derived risk scores was high, 0.96
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers tested whether OpenAI o1 could compute the PECARN intra-abdominal injury risk score for children after blunt abdominal trauma by reading physician notes written in free text. They compared its output against scores derived by clinicians, using records from two urban pediatric emergency departments.

The result suggests that clinical prediction rules could be calculated automatically from documentation, easing the labor of retrospective audits of how such rules are applied. The evidence comes from a retrospective analysis of a small sample, so performance in prospective care or other settings remains untested.
limitation: The analysis was retrospective, drew on only two urban pediatric emergency departments, and included just 196 children with complete notes, so it does not show how the model performs prospectively or elsewhere.
tag: Evidence-backed gain
key_points: The study analyzed 196 children with complete ED notes after blunt abdominal trauma. | Binary risk scores reached 94.4% sensitivity and 98.2% specificity against clinician-derived standards. | Multicategorical agreement by weighted Cohen's kappa was 0.96. | Extraction of individual PECARN predictor variables had accuracy above 94%.
rundown: The PECARN rule identifies children at very low risk for clinically significant intra-abdominal injury after blunt abdominal trauma. Evaluating whether clinicians use it in practice normally requires resource-intensive chart review, which motivated the test of an automated approach.

The study was a retrospective secondary analysis at two urban pediatric emergency departments. OpenAI o1 was compared with clinician-derived standards both for the final risk scores and for extracting each predictor variable from unstructured physician notes.

Reported accuracy was 95.4% with an F1 score of 0.97. The authors conclude that the LLM performed comparably to clinicians and that LLMs may enable automated calculation of clinical prediction rules from clinical documentation.
sources:
- peer_reviewed | JACEP Open | https://doi.org/10.1016/j.acepjo.2026.100525 | 2026-10-08
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

66431d0f2c2324217ab3a4245abb13604a2a39cfac851421eebb14c1e5b91e54

previous

0000000000000000000000000000000000000000000000000000000000000000