Real-Time Acute Kidney Injury Perioperative Prediction Clinical Trial (ML-AKI)

University of California San Francisco (UCSF)

Status: Begins enrollment in 4 months

Conditions: Surgery Complications; Acute Kidney Injury; Anesthesia

Treatments: Device: EHR-Embedded AKI Risk Score; Device: EHR-Embedded AKI Risk Score with Best Practice Advisory

Study type: Interventional

Funder types: Other; NIH

Identifiers

NCT07604662

22-37005

K23GM151611-03 (U.S. NIH Grant/Contract)

Details and patient eligibility

About

This investigator-initiated, pragmatic trial evaluates whether displaying a machine learning (ML)- derived perioperative AKI risk score-alone or paired with an interruptive Best/Our Practice Advisory (BPA/OPA)-improves kidney-protective care and reduces kidney injury after non-obstetric surgery at UCSF. Approximately 75-100 attending anesthesiologists (clusters) are randomized 1:1:1 to: (a) Control (risk score hidden), (b) Score Only (visible preoperative AKI risk probability with passive KDIGO bundle recommendation), or (c) Score + BPA (visible risk plus interruptive KDIGO prompt for high-risk patients). CRNAs/residents follow their attending' s assignment. Adult inpatients (age ≥18) with expected overnight stay and eGFR ≥15 mL/min/1.73 m² are included; obstetrics, chronic dialysis, and kidney transplant patients are excluded. The underlying preoperative model was prospectively validated at UCSF and outperforms anesthesiologist risk estimation reported in the literature. The model was reviewed and approved by the AI Oversight Committee at UCSF. Primary endpoint is the continuous change in serum creatinine (mg/dL) from baseline to POD 1-2. Secondary outcomes include KDIGO-defined AKI, adherence to bundle elements (hemodynamics, balanced fluids, nephrotoxin avoidance, glycemic control), intraoperative hypotension time, fluid volumes, nephrotoxin exposure, perioperative hyperglycemia, length of stay, unplanned ICU transfer, readmission, dialysis, and in-hospital mortality. Data are obtained from the EHR; analysts are blinded. No direct subject interaction is planned; the investigators will request a waiver of patient consent. The study aims to demonstrate that ML-enabled, workflow-embedded decision support can safely and feasibly improve guideline concordant care and decrease early postoperative kidney injury.

Enrollment: 25,518 estimated patients

Sex: All

Ages: 18+ years old

Volunteers: Accepts Healthy Volunteers

Inclusion criteria

• Adults ≥18 years undergoing non-obstetric surgery at UCSF.
• Inpatient cases with expected overnight stay.
• Baseline eGFR ≥15 mL/min/1.73 m².
• Managed by an attending anesthesiologist randomized to one of three arms (CRNAs/residents follow attending).
• Data available in the UCSF EHR for risk scoring and outcomes.

Exclusion criteria

• Obstetric procedures.
• Chronic dialysis patients.
• Kidney transplant recipients.
• Cases without baseline creatinine/eGFR or missing essential EHR elements needed for scoring/outcomes (operational exclusions).
• Outpatient procedures without expected overnight stay.

Trial design

Primary purpose: Screening

Allocation: Randomized

Interventional model: Parallel Assignment

Masking: None (Open label)

---

25,518 participants in 3 patient groups

Control Arm

No Intervention group

Description:

Participants receive usual perioperative care with a placeholder blank display without the machine learning-derived acute kidney injury (AKI) risk score. The clinical decision support tool remains hidden in the electronic health record, and no alerts or recommendations related to the study are shown.

Acute Kidney Injury Risk Score Only

Experimental group

Description:

A machine learning-derived preoperative AKI risk score is displayed within the electronic health record for high-risk patients. A passive recommendation indicating that the patient may benefit from a KDIGO-based kidney-protective bundle is provided. The information is advisory only, and no interruptive alerts are used.

Treatment:

Device: EHR-Embedded AKI Risk Score

Acute Kidney Injury Risk Score with Best Practice Advisory

Experimental group

Description:

The machine learning-derived AKI risk score is displayed within the electronic health record for high-risk patients, accompanied by an interruptive Best Practice Advisory (BPA) that notifies providers that the patient may benefit from a KDIGO-based kidney-protective bundle. The alert is advisory only and does not mandate clinical actions.

Treatment:

Device: EHR-Embedded AKI Risk Score with Best Practice Advisory