Predicting acute kidney injury onset using a random forest algorithm using electronic medical records of COVID-19 patients: the CRACoV-AKI model

Abstract

Introduction: Acute kidney injury (AKI) is a serious and common complication of SARS-CoV-2 infection. Most risk assessment tools for AKI have been developed in the intensive care unit or elderly population. As the COVID-19 pandemic is transitioning into an endemic state, there is an unmet need for prognostic scores tailored to this population.

Objectives: Development of a robust predictive model for the occurrence of AKI in hospitalized patients with COVID-19.

Patients and methods: Electronic medical records of all adult inpatients admitted between March 2020 and January 2022 were extracted from a large, tertiary care center with reference status in Lesser Poland. We screened 5806 patients with SARS-CoV-2 infection confirmed with polymerase chain reaction test. After excluding subjects with absent serum creatinine values or mild disease course (less than 7 days of inpatient care), 4630 patients were recruited. Data was randomly split into a training (N = 3462) and test (N = 1168) cohort. A random forest model was tuned with feature engineering based on expert advice and metrics evaluated in nested cross-validation to reduce bias.

Results: Nested cross-validation yielded an AUC (area under the curve) with a range of 0.793–0.807 and an average performance of 0.798. Model explanation techniques from a global perspective suggest respiratory support, chronic kidney disease and procalcitonin are among the most important variables in permutation tests.

Conclusions: The CRACoV-AKI model enables AKI risk stratification among hospitalized patients with COVID-19. Machine learning-based tools may thus offer additional decision-making support for specialist providers.