Machine learning approaches were introduced for better or comparable predictive ability than statistical analysis to predict postoperative outcomes. We sought to compare the performance of machine learning approaches with that of logistic regression analysis to predict acute kidney injury after cardiac surgery. We retrospectively reviewed 2010 patients who underwent open heart surgery and thoracic aortic surgery. Baseline medical condition, intraoperative anesthesia, and surgery-related data were obtained. The primary outcome was postoperative acute kidney injury (AKI) defined according to the Kidney Disease Improving Global Outcomes criteria. The following machine learning techniques were used: decision tree, random forest, extreme gradient boosting, support vector machine, neural network classifier, and deep learning. The performance of these techniques was compared with that of logistic regression analysis regarding the area under the receiver-operating characteristic curve (AUC). During the first postoperative week, AKI occurred in 770 patients (38.3%). The best performance regarding AUC was achieved by the gradient boosting machine to predict the AKI of all stages (0.78, 95% confidence interval (CI) 0.75–0.80) or stage 2 or 3 AKI. The AUC of logistic regression analysis was 0.69 (95% CI 0.66–0.72). Decision tree, random forest, and support vector machine showed similar performance to logistic regression. In our comprehensive comparison of machine learning approaches with logistic regression analysis, gradient boosting technique showed the best performance with the highest AUC and lower error rate. We developed an Internet–based risk estimator which could be used for real-time processing of patient data to estimate the risk of AKI at the end of surgery.

The cDNAs and genes encoding the intron lariat-debranching enzyme were isolated from the nematode Caenorhabditis elegans and the fission yeast Schizosaccharomyces pombe based on their homology with the Saccharomyces cerevisiae gene. The cDNAs were shown to be functional in an interspecific complementation experiment; they can complement an S. cerevisiae dbr1 null mutant. About 2.5% of budding yeast S. cerevisiae genes have introns, and the accumulation of excised introns in a dbr1 null mutant has little effect on cell growth. In contrast, many S. pombe genes contain introns, and often multiple introns per gene, so that S. pombe is estimated to contain ϳ40 times as many introns as S. cerevisiae. The S. pombe dbr1 gene was disrupted and shown to be nonessential. Like the S. cerevisiae mutant, the S. pombe null mutant accumulated introns to high levels, indicating that intron lariat debranching represents a rate-limiting step in intron degradation in both species. Unlike the S. cerevisiae mutant, the S. pombe dbr1::leu1 ؉ mutant had a severe growth defect and exhibited an aberrant elongated cell shape in addition to an intron accumulation phenotype. The growth defect of the S. pombe dbr1::leu1؉ strain suggests that debranching activity is critical for efficient intron RNA degradation and that blocking this pathway interferes with cell growth.Pre-mRNA introns are excised from precursor RNA in the form of a lariat RNA structure during the process of RNA splicing. In this structure, the 5Ј end of the intron is joined via a 2Ј-5Ј phosphodiester linkage to an internal adenosine (A) residue (the branchpoint nucleotide). These introns are excised by a two-step reaction (9,25,29). In the first step, the 5Ј splice site is cleaved to yield a 5Ј exon intermediate (exon 1). Concurrently, the 5Ј end of the intron is joined to the branch site (usually an adenosine residue), forming a 2Ј-5Ј phosphodiester bond. This forms the lariat intermediate RNA (or 2/3 intermediate) consisting of a lariat-form intron joined to the 3Ј exon (exon 2). The exon 1 and lariat intron-exon 2 intermediates are subsequently resolved by a second cleavage reaction at the 3Ј splice site which releases the mature lariat form of intron RNA and the ligation product of exons 1 and 2 via a 3Ј-5Ј phosphodiester bond, the mature mRNA. The excised intron lariat RNA is rapidly degraded in vivo, with a half-life of only a few seconds (31). Thus, in general, the spliced introns are not sufficiently stable in eukaryotic cells to be detected in vivo.In contrast to the splicing pathway leading from pre-mRNA to mRNA, very little is known about the pathway of intron degradation. Considering the enormous quantity of intron RNA that is discarded during the process of normal gene expression in mammalian cells, for example, intron degradation is likely to be an important pathway for normal cellular function. Only one component of the intron turnover pathway has been identified biochemically and genetically, the RNA lariat-debranching enzyme (DBR). DBR specifically hydrolyze...

The cDNA encoding the human RNA lariat debranching enzyme (hDBR1) was identified and cloned by searching the Expressed Sequence Tag (EST) database and screening a HeLa cDNA library, based on predicted amino acid sequence homologies with the Saccharomyces cerevisiae, Schizosaccharomyces pombe and Caenorhabditis elegans debranching enzymes. The hDBR1 cDNA expressed in Escherichia coli showed debranching activity in vitro and was also shown to be functional in an interspecies specific complementation experiment. hDBR1 cDNA in a S. cerevisiae expression vector complemented the intron accumulation phenotype of a S. cerevisiae dbr1 null mutant. Integration of the cDNA for hDBR1 into the ura4 locus of S. pombe also complemented both the intron accumulation and slow growth phenotypes of a S. pombe dbr1 null mutant strain. Comparison of the amino acid sequence of hDBR1 with the other DBR protein sequences showed several conserved regions, with 40, 44 and 43% identity to the S. cerevisiae, S. pombe and C. elegans debranching enzymes, respectively.

The effects of remote ischemic preconditioning (RIPC) in cardiac surgery have been inconsistent. We investigated whether anesthesia or beta-blockers interfere with RIPC cardioprotection. Fifty patients undergoing cardiac surgery were randomized to receive limb RIPC (four cycles of 5-min of upper arm cuff inflation/deflation) in the awake state (no-anesthesia; n = 17), or under sevoflurane (n = 17) or propofol (n = 16) anesthesia. In a separate crossover study, 11 healthy volunteers received either carvedilol or no medication prior to RIPC. Plasma dialysates were obtained and perfused through an isolated male Sprague–Dawley rat heart subjected to 30-min ischemia/60-min reperfusion, following which myocardial infarct (MI) size was determined. In the cardiac surgery study, pre-RIPC MI sizes were similar among the groups (39.7 ± 4.5% no-anesthesia, 38.9 ± 5.3% sevoflurane, and 38.6 ± 3.6% propofol). However, post-RIPC MI size was reduced in the no-anesthesia group (27.5 ± 8.0%; p < 0.001), but not in the anesthesia groups (35.7 ± 6.9% sevoflurane and 35.8 ± 5.8% propofol). In the healthy volunteer study, there was a reduction in MI size with RIPC in the no-carvedilol group (41.7 ± 4.3% to 30.6 ± 8.5%; p < 0.0001), but not in the carvedilol group (41.0 ± 4.0% to 39.6 ± 5.6%; p = 0.452). We found that the cardioprotective effects of limb RIPC were abolished under propofol or sevoflurane anesthesia and in the presence of carvedilol therapy.