Artificial neural network modeling to predict the plasma concentration of aminoglycosides in burn patients

Pharmacogenetic-guided Warfarin Dosing Algorithm in African-Americans

“…52 Artificial neural networks have been used extensively in medical and therapeutic applications. These applications include prediction of aminoglycosides plasma concentration in burn patients, 53 description of erythropoiesis kinetics in patients with renal failure, 54 screening for risk factors related to liver diseases, 55 and prediction of cyclosporine dose in patients after kidney transplantation. 56 Recently, a warfarin dosing algorithm in AfricanAmericans that incorporates rs12777823 genotype was developed by Hernandez et al 57 Compared with the present model, Hernandez et al incorporated more subjects to derive the algorithm (n = 349).…”

Machine learning in burn care and research: A systematic review of the literature

“…When ML algorithms were compared against traditional statistical methods such as linear modeling or logistic regression, the former always outperformed the latter [22,23,27,28]. For 3 studies employing and comparing multiple ML techniques, there was no consensus on which technique was the best [28,29,32].…”

“…In total 5105 patients with acute thermal injury, 171 clinical burn wounds, 180 9-mer peptides, and 424 12-mer peptides were included in the studies. The studies focused on burn diagnosis (burn depth and classification) [21,24,25,30,31] and prediction of different targets ranging from plasma concentration of aminoglycosides in burn patients [22,23], response of aminoglycoside antibiotics against methicillinresistant Staphylococcus aureus infection in patients with acute thermal injury [27], hospital LOS [19,28], survival/ mortality [19,20,29,32], burn healing time [26], and antimicrobial peptides with high levels of antimicrobial activity [33]. There were no disagreements between the two reviewers as to study inclusion or data end point analysis.…”

“…Although ML algorithms are often tested using a cross-validation approach, with ten-fold cross-validation being commonly applied during development, only 9 studies reported this. Of the 9 studies, 3 investigated algorithm performance by leave-one-out crossvalidation [22,23,27], 1 used a multi-objective cross-validation approach [32], and 3 studies [19,26,29] applied ten-fold cross-validation.…”

“…Of those, 785 focused on the wrong topic (e.g., robotics) and were excluded from the analysis. Afterwards, titles and abstracts of 722 articles were screened, and a total of 15 studies were identified and included in the analysis [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33]. 4 studies [21,24,25,30] used the same database of wound images, and 3 studies [22,23,27] used the same patient dataset.…”