EHR-Embedded Risk Score for Acute Kidney Injury

(ML-AKI Trial)

AB
Overseen ByAndrew Bishara, MD
Age: 18+
Sex: Any
Trial Phase: Academic
Sponsor: University of California, San Francisco
No Placebo GroupAll trial participants will receive the active study treatment (no placebo)

What You Need to Know Before You Apply

What is the purpose of this trial?

This trial aims to determine if a machine learning tool can reduce kidney problems after surgery. The tool, known as the EHR-Embedded AKI Risk Score, provides doctors with a risk score for acute kidney injury (AKI) and suggests kidney protection strategies. Some doctors will receive only the risk score, while others will also receive alerts prompting them to take action. This trial includes adults undergoing non-obstetric surgery at UCSF, who are staying overnight and have specific kidney function levels. For those concerned about kidney issues after surgery, this trial may help identify better kidney protection methods. As an unphased trial, it offers a unique opportunity to contribute to innovative research that could improve kidney protection strategies for future patients.

What prior data suggests that this machine learning-derived risk score is safe for improving kidney-protective care?

Research has shown that the EHR-Embedded AKI Risk Score tool is generally safe for patients. Integrated into electronic health records, it helps identify patients at high risk for acute kidney injury (AKI) without requiring extra blood tests. This tool enables doctors to focus on patients needing more attention, potentially avoiding unnecessary treatments.

The version with Best Practice Advisory includes alerts for doctors, offering suggestions to protect kidney health. These alerts serve as recommendations, allowing doctors to choose the best action for each patient.

Oversight committees have reviewed and approved both tools to ensure safety and ethics. Studies have not reported any significant harmful effects, making these tools a promising and safe option for improving patient care.12345

Why are researchers excited about this trial?

Researchers are excited about this trial because it explores the potential of using machine learning to predict acute kidney injury (AKI) risk in patients. Unlike traditional methods that rely on general clinical assessments, this trial tests an electronic health record (EHR)-embedded risk score, which is tailored to each patient's data. Additionally, one arm of the trial includes a Best Practice Advisory (BPA) that actively alerts healthcare providers about high-risk patients, suggesting a kidney-protective bundle based on KDIGO guidelines. This approach aims to provide personalized, proactive care, potentially reducing the incidence of AKI by catching risk factors early and guiding timely interventions.

What evidence suggests that this trial's treatments could be effective for reducing acute kidney injury?

This trial will evaluate the effectiveness of an EHR-Embedded AKI Risk Score in predicting acute kidney injury (AKI). Studies have shown that electronic health records (EHR) with a specialized computer program can predict AKI before it occurs. This tool identifies high-risk patients even without complete kidney health information, enabling doctors to take preventive steps sooner. Research indicates that these tools improve AKI prediction by 5% compared to traditional methods. In this trial, one arm will use the EHR-Embedded AKI Risk Score alone, while another will combine the Risk Score with a Best Practice Advisory (BPA), which provides clear guidance to doctors. This system can further increase the chances of preventing kidney damage after surgery. Overall, evidence supports using these digital tools to enhance patient care and reduce the risk of AKI.12346

Who Is on the Research Team?

AB

Andrew Bishara, MD

Principal Investigator

University of California, San Francisco

Are You a Good Fit for This Trial?

This trial is for adults (18 or older) having non-obstetric surgery at UCSF who are expected to stay overnight and have moderately functioning kidneys (eGFR ≥15). It excludes those on chronic dialysis, kidney transplant patients, and obstetrics cases.

Inclusion Criteria

* Data available in the UCSF EHR for risk scoring and outcomes.
* Inpatient cases with expected overnight stay.
* Baseline eGFR ≥15 mL/min/1.73 m².
See 2 more

Timeline for a Trial Participant

Screening

Participants are screened for eligibility to participate in the trial

2-4 weeks

Preoperative Assessment

Machine learning-derived AKI risk score is assessed preoperatively for high-risk patients

1 day

Intraoperative Monitoring

Monitoring of intraoperative parameters including fluid volumes, nephrotoxin exposure, and hypotension

Duration of surgery

Postoperative Monitoring

Monitoring of serum creatinine changes and other secondary outcomes from operation to postoperative day 7

7 days

Follow-up

Participants are monitored for safety and effectiveness after treatment, including dialysis requirement and hospital readmission

180 days

What Are the Treatments Tested in This Trial?

Interventions

  • EHR-Embedded AKI Risk Score
  • EHR-Embedded AKI Risk Score with Best Practice Advisory

Trial Overview

The study tests if showing anesthesiologists a machine learning-based risk score for acute kidney injury—either alone or with an alert—improves care and reduces kidney problems after surgery. Patients are grouped based on their anesthesiologist's assignment.

How Is the Trial Designed?

3

Treatment groups

Experimental Treatment

Active Control

Group I: Acute Kidney Injury Risk Score with Best Practice AdvisoryExperimental Treatment1 Intervention
Group II: Acute Kidney Injury Risk Score OnlyExperimental Treatment1 Intervention
Group III: Control ArmActive Control1 Intervention

Find a Clinic Near You

Who Is Running the Clinical Trial?

University of California, San Francisco

Lead Sponsor

Trials
2,636
Recruited
19,080,000+

National Institute of General Medical Sciences (NIGMS)

Collaborator

Trials
315
Recruited
251,000+

Citations

EHR-Embedded Risk Score for Acute Kidney Injury (ML- ...

* Data available in the UCSF EHR for risk scoring and outcomes. * Inpatient cases with expected overnight stay. * Baseline eGFR ≥15 mL/min ...

Electronic Health Record-Based Predictive Models for Acute ...

The AKI screening tool can be incorporated into EHR systems to identify high risk patients without serum creatinine data, enabling targeted laboratory testing.

Real-Time Acute Kidney Injury Perioperative Prediction ...

This investigator-initiated, pragmatic trial evaluates whether displaying a machine learning (ML)- derived perioperative AKI risk score-alone or paired with ...

Implementation of an Electronic Health Records–Based ...

In this study, we found that an EHR-based safe contrast limit tool using automatically derived patient data performed well in predicting CA-AKI ...

Clustering analysis of multi-site electronic health records ...

Despite its mild presentation, AKI-1 outcomes range from rapid renal recovery to prolonged kidney dysfunction. For instance, in a cohort of AKI ...

Reducing Postoperative Acute Kidney Injury: A Regional ...

With the full implementation of risk tools, alerts, and order sets in Phase 3, AKI incidence declined to 400 cases per month. Process control analysis showed ...