Artificial intelligence and machine learning for predicting acute kidney injury in severely burned patients: A proof of concept.

BACKGROUND: Burn critical care represents a high impact population that may benefit from artificial intelligence and machine learning (ML). Acute kidney injury (AKI) recognition in burn patients could be enhanced by ML. The goal of this study was to determine the theoretical performance of ML in augmenting AKI recognition. METHODS: We developed ML models using the k-nearest neighbor (k-NN) algorithm. The ML models were trained-tested with clinical laboratory data for 50 adult burn patients that had neutrophil gelatinase associated lipocalin (NGAL), urine output (UOP), creatinine, and N-terminal B-type natriuretic peptide (NT-proBNP) measured within the first 24 h of admission. RESULTS: Half of patients (50%) in the dataset experienced AKI within the first week following admission. ML models containing NGAL, creatinine, UOP, and NT-proBNP achieved 90-100% accuracy for identifying AKI. ML models containing only NT-proBNP and creatinine achieved 80-90% accuracy. Mean time-to-AKI recognition using UOP and/or creatinine alone was achieved within 42.7 ± 23.2 h post-admission vs. within 18.8 ± 8.1 h via the ML-algorithm. CONCLUSIONS: The performance of UOP and creatinine for predicting AKI could be enhanced by with a ML algorithm using a k-NN approach when NGAL is not available. Additional studies are needed to verify performance of ML for burn-related AKI.

High-Dose Ascorbic Acid for Burn Shock Resuscitation May Not Improve Outcomes.

High dose ascorbic acid (HDAA) has been touted to ameliorate inflammation and reduce fluid requirements during burn shock resuscitation (BSR). Whether this leads to improved outcomes is not known. The authors' aim for this study was to compare ventilator days, ventilator-associated pneumonia, and mortality between patients who did and did not receive HDAA during BSR.The authors performed a retrospective case control study from 2012 to 2015. They identified 38 patients (HDAA) who received HDAA during BSR. Using age and %TBSA, the authors identified and matched 42 control patients (CTL) who did not receive HDAA for BSR during that same time period. The authors collected data for age, %TBSA, hospital days (LOS), ventilator days (VENT), inhalation injury (INH), ventilator-associated pneumonia (VAP), and mortality (MORT).There were no differences in age and %TBSA or %TBSA of third-degree burn injury between groups. There was no significant difference in the incidence of INH (HDAA-52% vs CTL-36%, P = .17) and the groups had similar LOS and VENT. Additionally, there was no significant difference in VAP incidence (HDAA-29% vs CTL-14%, P = .13) or mortality (HDAA-26% vs CTL-23%, P = .8). HDAA patients had a numerically higher incidence of acute renal failure requiring dialysis (23 vs 7%, P = .06) which was confirmed in a multivariate analysis (odds ratio 5.4; 95% confidence interval 1.1-26). HDAA, while potentially reducing inflammation and fluid requirements during BSR, may not improve any meaningful outcomes such as ventilator requirements, ventilator-associated pneumonia, and mortality.

Development of novel smart device based application for serial wound imaging and management.

Burn wound photography has diverse clinical applications; however, inherent technological limitations mitigate its utility. Limitations include lack of quality control, serial imaging, complexity, and expense. With the performance gap between smart devices and digital cameras rapidly narrowing, and computing performance increasing, smart devices are poised to uniquely address these limitations and enhance the field of wound photography. To this end, we developed a proof-of-concept smart device application addressing the limitations of traditional photography and meeting the needs of burn clinicians. The result was an innovative smart device application providing user-friendly serial imaging and informatics capabilities at the patient bedside. The application generated images with significantly higher brightness (2.4±1.07 vs. 3.8±1.69, n=15, p<0.05) and higher contrast (255±0.00 vs. 236.3±5.64, n=15, p<0.0001), more consistent positioning (1.22±0.03 vs. 2.08±0.61, n=15, p<0.0001) and zoom (18.14 vs. 14.29, n=15, p<0.0001) compared to those taken by a basic compact digital camera using default settings. Surveyed clinician end-users reported greater functionality (20 vs. 0, n=20, p<0.001), and a more intuitive interface (18 vs. 2, n=20, p<0.001) with the application. We report consistent serial wound imaging and informatics are both feasible on a smart device platform. These findings will pave the way for new smart device-based clinical applications.