Convolutional Neural Network Model for Intensive Care Unit Acute Kidney Injury Prediction

INTRODUCTION

Acute kidney injury (AKI) is common among hospitalized patients and has a significant impact on morbidity and mortality. Although early prediction of AKI has the potential to reduce adverse patient outcomes, it remains a difficult condition to predict and diagnose. The purpose of this study was to evaluate the ability of a machine learning algorithm to predict for AKI as defined by Kidney Disease: Improving Global Outcomes (KDIGO) stage 2 or 3 up to 48 hours in advance of onset using convolutional neural networks (CNNs) and patient electronic health record (EHR) data.

METHODS

A CNN prediction system was developed to use EHR data gathered during patients' stays to predict AKI up to 48 hours before onset. A total of 12,347 patient encounters were retrospectively analyzed from the Medical Information Mart for Intensive Care III (MIMIC-III) database. An XGBoost AKI prediction model and the sequential organ failure assessment (SOFA) scoring system were used as comparators. The outcome was AKI onset. The model was trained on routinely collected patient EHR data. Measurements included area under the receiver operating characteristic (AUROC) curve, positive predictive value (PPV), and a battery of additional performance metrics for advance prediction of AKI onset.

RESULTS

On a hold-out test set, the algorithm attained an AUROC of 0.86 and PPV of 0.24, relative to a cohort AKI prevalence of 7.62%, for long-horizon AKI prediction at a 48-hour window before onset.

CONCLUSION

A CNN machine learning-based AKI prediction model outperforms XGBoost and the SOFA scoring system, revealing superior performance in predicting AKI 48 hours before onset, without reliance on serum creatinine (SCr) measurements.

Neutrophil gelatinase-associated lipocalin (NGAL) fails as an early predictor of contrast induced nephropathy in chronic kidney disease (ANTI-CI-AKI study)

The aim of the study was to evaluate the diagnostic accuracy of urinary neutrophil gelatinase- associated lipocalin (uNGAL) in patients with chronic kidney disease (CKD) as an early biomarker for contrast induced acute kidney injury (CI-AKI) and to investigate whether patients with an uNGAL increase might benefit from an additional intravenous volume expansion with regard to CI-AKI-incidence. We performed a prospective randomized controlled trial in 617 CKD-patients undergoing intra-arterial angiography. Urinary NGAL was measured the day before and 4-6hrs after angiography. In the event of a significant rise of uNGAL patients were randomized either into Group A, who received intravenous saline post procedure or Group B, who did not receive post-procedural i.v. fluids. Ten patients (1.62%) exhibited a significant rise of uNGAL after angiography and were randomized of whom one developed a CI-AKI. In the entire cohort the incidence of CI-AKI was 9.4% (58 patients) resulting in a specificity of 98.4% (95% CI: 97.0-99.3%) and a sensitivity of 1.72% (95% CI: 0.044-9.2%) of uNGAL for the diagnosis of CI-AKI. In this study uNGAL failed to predict CI-AKI and was an inadequate triage tool to guide an early intervention strategy to prevent CI-AKI.

CLINICAL TRIAL REGISTRATION

URL: http://www.clinicaltrials.gov. Unique identifier: NCT01292317.

Acute kidney injury clinical trial design: old problems, new strategies

Apart from supportive dialysis there are no universally accepted interventions in acute kidney injury (AKI). We have summarized the outcomes of all published randomized, placebo-controlled studies of non-dialysis treatment of AKI. Forty-nine trials were identified, only one of which was in a paediatric population. Sixteen trials had positive outcomes; these trials are not comparable in terms of methodology used or outcomes assessed, and they share many of the problems of the negative trials. We discuss the flaws in clinical trial design that have contributed to poor or uncertain outcomes and propose minimum requirements for future trials. In particular, future trials should incorporate biomarkers specific to the etiology of the AKI, and treatment should match the phase of injury.

Challenges facing early detection of acute kidney injury in the critically ill

Recent advances in the detection of acute kidney injury (AKI) afford the possibility of early intervention. Proteomics and genomics have identified many markers of tubular cell injury, some of which are manifest in the urine. One trial has used novel injury biomarkers to recruit patients to an intervention prior to an elevation in plasma creatinine. This trial and other recent studies have shown that the use of biomarkers of injury will depend on the time the patient presents following insult to the kidney, the likely cause of that insult, and the pre-injury renal function of that patient. The definition of AKI is likely to change in the near future to include a measure of injury. We anticipate novel therapies becoming available following successful trials that utilize the methodology of early intervention following an elevated injury biomarker.