medRxiv preprint: automated SOFA reconstruction improved respiratory variable recovery from 33% to 100%.
The study focuses on automated reconstruction of sepsis severity from unstructured clinical notes. In a MIMIC-III cohort of n=284, the approach improved respiratory variable recovery from 33% to 100% and reclassified critical SOFA ≥ 10 cases from 29% to 61%.
Because this is a preprint under journal review, the listing keeps the status visible alongside the source link and headline metrics.
The headline numbers from the paper, in plain language.
How the model performed
Higher is better
Data: Up from 33% in MIMIC-III notes (n=284) · Up from 29% — medRxiv 2025
| Metric | Value | What it means |
|---|---|---|
| Respiratory variable recovery | 100% | Up from 33% in MIMIC-III notes (n=284) |
| Critical cases reclassified (SOFA ≥ 10) | 61% | Up from 29% |
Katherine Monsalve Barrientos, Natalia Castaño-Villegas, Elmer Escandón, Jose Zea, Laura Velásquez. From free text to SOFA score: automated reconstruction of sepsis severity from unstructured clinical notes. medRxiv. 2025. https://medrxiv.org/cgi/content/short/2025.12.17.25342509v1
@article{barrientos2025from,
title = {{From free text to SOFA score: automated reconstruction of sepsis severity from unstructured clinical notes}},
author = {Katherine Monsalve Barrientos and Natalia Castaño-Villegas and Elmer Escandón and Jose Zea and Laura Velásquez},
journal = {medRxiv},
year = {2025},
url = {https://medrxiv.org/cgi/content/short/2025.12.17.25342509v1},
note = {Preprint}
}
1 min read
1 min read
1 min read
For hospitals and insurers
We run the model on your own clinical records with your team, under your privacy and governance rules.
Start a research partnership