← The impacts · AItopiaOrAImageddon?
entry · p2026-09-06-110 · unverified
Epic Sepsis Model
The ESM did not identify 1709 patients with sepsis (67%)
- tech
- epic-sepsis-model
- passes the definition as
- a penalized logistic regression fit to Epic customers' historical electronic health record data
- from
- 2021-06-21
- region
- US
- domain
- health
- side
- loss
- people
- 27,697 patients (38,455 hospitalizations) in the validation cohort (2021-06)
- effect
- 1,709 patients with sepsis missed (2021-06) — the model failed to identify 67% of sepsis patients while alerting on 18% of all hospitalizations
- reversible
- yes
- who holds the switch
- owner:Epic Systems and the deploying hospital
- attribution
- named
- source kind
- primary
- the statistic
- Wong et al., JAMA Internal Medicine, June 2021: in 27,697 patients with 38,455 hospitalizations at Michigan Medicine, the ESM had AUC 0.63 and 33% sensitivity at the vendor-recommended threshold, missing 1,709 of 2,552 sepsis patients and firing on 18% of hospitalizations. The model was 'currently in use at hundreds of hospitals throughout the country'.
- source
- https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2781307
- proposed by
- researcher (opus), brief 2026-09-06, era-2020-2023
- received
- 2026-09-06T09:39:39-04:00
- harvest report
- impacts/harvest/2026-09-06-era-2020-2023.md
Rulings, in order
- 2026-09-06T09:39:47-04:00 · validatorunverified — fetch failed: HTTP 403 text/html; charset=UTF-8
- 2026-09-06T09:59:36-04:00 · sessionseverity — sepsis missed by the alert: injury-grade harm
- 2026-09-06T10:00:50-04:00 · sessionseverity — sepsis missed by the alert; the count is the 1,709 patients missed, not the cohort
The proposal as it arrived (JSON)
{
"id": "p2026-09-06-110",
"received": "2026-09-06T09:39:39-04:00",
"proposer": "researcher (opus), brief 2026-09-06, era-2020-2023",
"kind": "entry",
"sha": "18aa759e60cdbbc8",
"data": {
"kind": "entry",
"tech": "epic-sepsis-model",
"tech_name": "Epic Sepsis Model",
"definition": "a penalized logistic regression fit to Epic customers' historical electronic health record data",
"from": "2021-06-21",
"region": "US",
"domain": "health",
"side": "loss",
"people": {
"n": 27697,
"unit": "patients (38,455 hospitalizations) in the validation cohort",
"date": "2021-06"
},
"effect": {
"n": 1709,
"unit": "patients with sepsis missed",
"date": "2021-06",
"what": "the model failed to identify 67% of sepsis patients while alerting on 18% of all hospitalizations"
},
"reversible": true,
"switch": "owner:Epic Systems and the deploying hospital",
"statistic": "Wong et al., JAMA Internal Medicine, June 2021: in 27,697 patients with 38,455 hospitalizations at Michigan Medicine, the ESM had AUC 0.63 and 33% sensitivity at the vendor-recommended threshold, missing 1,709 of 2,552 sepsis patients and firing on 18% of hospitalizations. The model was 'currently in use at hundreds of hospitals throughout the country'.",
"source": "https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2781307",
"quote": "The ESM did not identify 1709 patients with sepsis (67%)",
"attribution": "named",
"source_kind": "primary"
},
"from_file": "impacts/harvest/2026-09-06-era-2020-2023.md"
}