Combining Biomarkers and Electronic Risk Scores to Predict AKI in Hospitalized Patients

Overview

The study's objective is to evaluate the additive value of renal biomarkers (from blood and urine) for identifying individuals at high risk for severe acute kidney injury (AKI) above that of a novel natural language processing (NLP)-based AKI risk algorithm. The risk algorithm is based on electronic health records (EHR) data (labs, vitals, clinical notes, and test reports). Patients will enroll at the University of Chicago Medical Center and the University of Wisconsin Hospital, where the risk score will run in real time. The risk score will identify those patients with the highest risk for the future development of Stage 2 AKI and collect blood and urine for biomarker measurement over the subsequent 3 days.

Detailed description

The investigators hypothesize that combining the biomarkers with electronic health risk score will impact improvement in AKI risk stratification. Using a real time, externally validated electronic health record based AKI risk score, the investigators will enroll patients who are at high risk for the impending development of KDIGO Stage 2 AKI (top 10% of risk). Once identified and enrolled, patients will have blood and urine samples collected over the next 3 days. The investigators will recruit two cohorts of 400 patients across the two institutions. In the development cohort, the investigators will see if adding urinary or blood biomarkers of AKI can improve the ability of EHR-risk score to predict the development of Stage 2 AKI and other outcomes. The investigators will compare the area under the receiver operator characteristic curve (AUC) for the risk score alone versus the risk score plus biomarkers. The investigators will then seek to validate our findings in a separate cohort of 400 patients.

Interventions

• Device ESTOP - AKI 2.0
  Medical software as a Noninvasive medical device, which at the time of the project will not implement directly into subject/clinical care.

Primary outcome measures

• Developing KDIGO stage 2 AKI [Time frame: Within 7 days of enrollment]

Secondary outcome measures (5)

• Development of KDIGO stage 3 AKI [Time frame: within 12 hour of each observation, within 7 days of enrollment and 90 day MAKE outcome]
• Recipient of renal replacement therapy(RRT) [Time frame: within 7 days of enrollment and 90 day make outcome]
• Clinical indication for the receipt of renal replacement therapy(RRT) [Time frame: within 12 hour of each observation, within 7 days of enrollment and 90 day make outcome]
• Change in Mortality Status during hospitalization [Time frame: within 12 hour of each observation, within 7 days of enrollment and during current hospitalization]
• Major Adverse Kidney Events (MAKE) Outcomes [Time frame: 3 months (90 days)]

Eligibility criteria

Inclusion criteria

• Age ≥ 18 years
• E-STOP AKI 2.0 score in the top 10% of risk (historically from all hospitalized patients) within the last 12 hours. (First time across this 10% risk threshold during this hospital stay).
• Admitted to an inpatient ward, intermediate, or ICU care at the University of Chicago Medical Center (UCMC) or University of Wisconsin Health (UWHealth). (No Emergency Department patients)
• Patient or their legally authorized representative must be able to read, speak, and understand English, for the purposes of consenting. Otherwise, inclusion in this protocol will be done without regard to race, ethnic origin or gender

Exclusion criteria

• Voluntary refusal or missing written consent of the patient / legal representative.
• Patients with a known history of end-stage renal disease on dialysis (including renal transplantation).
• Patients without a measured serum creatinine value during their inpatient stay.
• Patients with a creatinine >4.0 mg/dl at the time of admission or available in the EHR from the last 6 months
• Patients with prior episode of KDIGO defined AKI during this same hospitalization- regardless of E-STOP AKI 2.0 score
• Patients with prior renal consultation during their admission.
• Patient with an E-STOP AKI 2.0 above the top 10% risk threshold more than 12 hours ago during this same hospital stay.
• Incarcerated patients
• Pregnant patients

Criteria are shown verbatim from the registry (in English). Final eligibility is always assessed by the study center.

Healthy volunteers: No

Study design

Observational model

Cohort

Study locations

United States · 2 centers

• University of Chicago Medical Center — Chicago
• University of Wisconsin Hospital — Madison

Publications

• Koyner JL, Martin J, Carey KA, Caskey J, Edelson DP, Mayampurath A, Dligach D, Afshar M, Churpek MM. Multicenter Development and Validation of a Multimodal Deep Learning Model to Predict Moderate to Severe AKI. Clin J Am Soc Nephrol. 2025 Apr 15;20(6):766-778. doi: 10.2215/CJN.0000000695. PMID 40232856

Identifiers

NCT: NCT05988658 · IRB23-0343 · R01DK126933

Primary sources (government registries)

• ClinicalTrials.gov (US NIH)

View this study on ClinicalTrials.gov ↗