Early Recognition of Burn- and Trauma-Related Acute Kidney Injury: A Pilot Comparison of Machine Learning Techniques

Abstract

Severely burned and non-burned trauma patients are at risk for acute kidney injury (AKi). the study objective was to assess the theoretical performance of artificial intelligence (AI)/machine learning (ML) algorithms to augment AKi recognition using the novel biomarker, neutrophil gelatinase associated lipocalin (nGAL), combined with contemporary biomarkers such as n-terminal pro B-type natriuretic peptide (nt-proBnp), urine output (Uop), and plasma creatinine. Machine learning approaches including logistic regression (LR), k-nearest neighbor (k-nn), support vector machine (SVM), random forest (RF), and deep neural networks (DNN) were used in this study. The AI/ML algorithm helped predict AKI 61.8 (32.5) hours faster than the Kidney Disease and Improving Global Disease Outcomes (KDiGo) criteria for burn and non-burned trauma patients. nGAL was analytically superior to traditional AKi biomarkers such as creatinine and Uop. With ML, the AKi predictive capability of nGAL was further enhanced when combined with NT-proBNP or creatinine. The use of AI/ML could be employed with nGAL to accelerate detection of AKi in at-risk burn and non-burned trauma patients. Acute kidney injury (AKI) is a common complication among critically ill patients 1-4. Severely burned patients, in particular, have been shown to be at high-risk with up to 58% experiencing AKI 3-5. The early recognition of AKI helps guide fluid resuscitation and titrate dosing of nephrotoxic drugs in these populations. Unfortunately, traditional biomarkers of renal function such as creatinine and urine output (UOP) have been shown to be suboptimal at predicting AKI 6,7. Novel AKI biomarkers have been proposed, but widespread use in the United States remains limited. Advances in computational technology have rapidly facilitated the growth of artificial intelligence (AI) and machine learning (ML) 8,9. Studies have reported AI/ML aiding in the diagnosis of several disease and perhaps augment the performance of existing tests with varying degrees of success 10-13. Interestingly, recent investigations postulated AI/ML using a k-nearest neighbor (k-NN) approach could augment the identification of AKI in burn patients using only plasma creatinine, UOP and N-terminal pro-B-type natriuretic peptide (NT-proBNP) 14. Notably, that study was limited to burn patients-raising the question if these algorithms could apply to other critically ill populations and if k-NN was the optimal ML technique for AKI prediction. Severely burned patients have been shown to be fundamentally different from traditional non-burned trauma populations 15,16. Interestingly, AKI classification remains the same between both populations and based on the Kidney Disease and Improving Global Outcomes (KDIGO) criteria 17. This similarity offers a unique opportunity to determine if ML models developed in one population (i.e., burn patients) could be translated to another (i.e., non-burned trauma patients) and how KDIGO performs against such ML techniques. Notably, the KDIGO criteria relies solely on UOP and...

Physiologic risk factors for early acute kidney injury in severely injured patients

“…In victims of trauma, elevated plasma levels of NGAL were associated with a risk of the development of posttraumatic multiorgan dysfunction syndrome (14). Plasma levels of NGAL were superior to traditional AKI biomarkers such as: creatinine and UOP in AKI prediction both in burn and non-burned trauma patients and NGAL-based algorithm helped predict AKI 61.8 (32.5) hours faster than the Kidney Disease and Improving Global Disease Outcomes (KDIGO) criteria (15). In victims of trauma, urinary levels of AKI biomarkers (including NGAL) were associated with the injury severity, and early urinary NGAL levels correlated with subsequent AKI development, a need for renal replacement therapy, and mortality (9,16,17).…”

Novel application of an automated-machine learning development tool for predicting burn sepsis: proof of concept

“…Studies have indicated traditional sepsis criteria are not suitable for this unique population 1 . Recent investigations suggest ML may be able to identify unique pathologic patterns not recognized by the "human eye" and enhance the performance of traditional biomarkers for certain diseases (e.g., acute kidney injury) 8,11 . In this article, we report the use of ML for predicting sepsis in this unique high-risk burn population.…”

“…Machine learning sepsis algorithms were first developed using our exhaustive “traditional” non-automated ML approach 8 , 11 . The process entailed manually selecting various feature set combinations aided with an unsupervised select percentile techniques such as ANOVA F-classification for feature selection from the original dataset followed by building a large number of models on various supervised algorithms.…”

Artificial Intelligence in Acute Kidney Injury Risk Prediction

“…Rashidi and colleagues developed, internally validated, and compared ML models for early recognition of AKI in 50 burn and 51 trauma patients, including NGAL, NT-proBNP, SCr, and UO into the predictive model [119]. Their models were able to accurately predict AKI 62 h in advance [119].…”