Moderate evidence
Published recordHealthGood Space
More
Interpretable Recalibration of a Sequential Organ Failure Assessment-Derived Score Using Machine Learning-Guided Optimization: Multicenter Cohort Study
- Truvace record
- TRV-2026-2936raw data (JSON)
- Version
- v1Published
- First published
- Evidence type
- Observational study
- Sources
- 1 sourcePeer-reviewed
- sha256
- 6a506f7b8c40…check it
Background The Sequential Organ Failure Assessment (SOFA) score is central to risk stratification in intensive care units (ICUs), but its thresholds and 0- to 4-point increments were set by expert consensus in 1996 and never reestimated from outcome data. Machine learning models discriminate better while discarding the additive, point-based structure that makes SOFA auditable at the bedside, and methods for updating a legacy score rather than replacing it remain scarce. Objective This study aimed to develop and…
The gain
In a retrospective study of 14,848 first ICU admissions across 208 US hospitals, a machine learning-recalibrated SOFA-derived score predicted ICU death with an AUROC of 0.811 versus 0.751 for the reference score.
What this doesn’t fix
The score was only internally validated on a retrospective US cohort in which 90% of encounters with outcome data were excluded for missing measurements; external and prospective validation is required before clinical use.
Sources
1 on file- JMIR Medical Informaticsdoi:10.2196/99361
Cite this record
v1Truvace Impact Record TRV-2026-2936, v1: “Interpretable Recalibration of a Sequential Organ Failure Assessment-Derived Score Using Machine Learning-Guided Optimization: Multicenter Cohort Study.” Truvace, 2026-11-15. /record/TRV-2026-2936 (accessed on the day you copy it). sha256 6a506f7b8c400ea0…
Change history
1 versionEvery change made to this record since it was first published, and why. Nothing has changed since it went up.
Certified into the record
Check this record hasn’t been altered
How to check it yourself, without trusting this page
Download any version from /api/record/TRV-2026-2936 and hash its canonical field yourself, for example with shasum -a 256. You should get its content_hash. Each version’s text also ends with prev:and the hash of the version before it (version 1 points to 64 zeros). If anyone had changed a single character of any version after it was saved, those hashes wouldn’t line up.
ace