Research2026-08-27

Researchers built a continuously updating sepsis severity score from routine hospital chart data, trained on records from two US hospital systems. Patients who later died scored consistently higher than survivors across comparable baseline groups, and the score tracked changes in lactate within individual patients. The work is retrospective and the authors present it only as possible decision support alongside clinical judgement.

What changed

Sepsis severity scores in routine use rely on fixed variables and weights set decades ago.

What it unlocks

An hourly, continuously updating severity score for sepsis patients learned from patient records rather than fixed rules.

  • 29,116 and 7,691 patients, two sites
  • 1.19-1.64 points higher for non-survivors
  • 43 charted variables over 72 hours

What you need to act on it

  • retrospective research only, not deployed or cleared for clinical use

Sources

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.