Predicting In-Hospital Mortality at Admission to the Medical Ward: A Big-Data Machine Learning Model

BackgroundGeneral medical wards admit high-risk patients. Artificial intelligence algorithms can use big data for developing models to assess patients’ risk stratification. The aim of this study was to develop a mortality prediction machine learning model using data available at the time of admission to the medical ward.MethodsWe included consecutive patients (ages 18-100) admitted to medical wards at a single medical center (January 1, 2013-December 31, 2018). We constructed a machine learning model using patient characteristics, comorbidities, laboratory tests, and patients’ emergency department (ED) management. The model was trained on data from the years 2013 to 2017 and validated on data from the year 2018. The area under the curve (AUC) for mortality prediction was used as an outcome metric. Youden index was used to find an optimal sensitivity-specificity cutoff point.ResultsOf the 118,262 patients admitted to the medical ward, 6311 died (5.3%). The single variables with the highest AUCs were medications administered in the ED (AUC = 0.74), ED diagnosis (AUC = 0.74), and albumin (AUC = 0.73). The machine learning model yielded an AUC of 0.924 (95% confidence interval [CI]: 0.917-0.930). For Youden index, a sensitivity of 0.88 (95% CI: 0.86-0.89) and specificity of 0.83 (95% CI: 0.83–0.83) were observed. This corresponds to a false-positive rate of 1:5.9 and negative predictive value of 0.99.ConclusionA machine learning model outperforms single variables predictions of in-hospital mortality at the time of admission to the medical ward. Such a decision support tool has the potential to augment clinical decision-making regarding level of care needed for admitted patients.     

Calcium,mitochondria and the initiation of acute pancreatitis

Acute pancreatitis is characterized by necrosis of its parenchymal cells and influx and activation of inflammatory cells that further promote injury and necrosis. This review is intended to discuss the central role of disorders of calcium metabolism and mitochondrial dysfunction in the mechanism of pancreatitis development. The disorders are placed in context of calcium and mitochondria in physiologic function of the pancreas. Moreover, we discuss potential therapeutics for preventing pathologic calcium signals that injure mitochondria and interventions that promote the removal of injured mitochondria and regenerate new and heathy populations of mitochondria.     

Endoplasmic stress-inducing variants in carboxyl ester lipase and pancreatic cancer risk

BackgroundEndoplasmic reticulum (ER) stress-inducing variants in several pancreatic secretory enzymes have been associated with pancreatic disease. Multiple variants in CEL, encoding carboxyl ester lipase, are known to cause maturity-onset diabetes of the young (MODY8) but have not been implicated in pancreatic cancer risk.MethodsThe prevalence of ER stress-inducing variants in the CEL gene was compared among pancreatic cancer cases vs. controls. Variants were identified by next-generation sequencing and confirmed by Sanger sequencing. Variants of uncertain significance (VUS) were assessed for their effect on the secretion of CEL protein and variants with reduced protein secretion were evaluated to determine if they induced endoplasmic reticulum stress.ResultsER stress-inducing CEL variants were found in 34 of 986 cases with sporadic pancreatic ductal adenocarcinoma, and 21 of 1045 controls (P = 0.055). Most of the variants were either the CEL-HYB1 variant, the I488T variant, or the combined CEL-HYB1/I488T variant; one case had a MODY8 variant.ConclusionThis case/control analysis finds ER stress-inducing CEL variants are not associated with an increased likelihood of having pancreatic cancer.     

Serum lipase on postoperative day one is a strong predictor of clinically relevant pancreatic fistula after pancreaticoduodenectomy: A retrospective cohort

BackgroundIncreased postoperative serum amylase has been recently reported to be associated with increased postoperative morbidity, but studies on postoperative serum lipase are limited. The aim of this study was to evaluate the value of postoperative serum lipase in predicting clinically relevant postoperative pancreatic fistula (CR-POPF) after pancreaticoduodenectomy (PD).MethodA retrospective analysis was performed on 212 patients who underwent PD from September 2018 and March 2021, focusing on the association between postoperative day (POD) 1 serum lipase and CR-POPF.ResultsOverall, 108 (50.9%) patients had elevated serum lipase levels (>68 U/L) on POD 1. Patients with elevated serum lipase exhibited a significantly higher incidence of CR-POPF (37.0% vs. 6.7%, p < 0.001). Receiver operating characteristic (ROC) analyses showed improved diagnostic accuracy for POD 1 serum lipase compared with POD 1 serum amylase in predicting CR-POPF (AUC: 0.801 vs. 0.745, p = 0.029). Elevated serum lipase on POD 1 and elevated serum CRP on POD 3 were identified as independent predictors of CR-POPF. A simple early postoperative model, consisting of POD 1 serum lipase levels and POD 3 serum CRP levels, showed good discrimination (AUC 0.76, 95% CI 0.69–0.83) to identify the onset of CR-POPF.ConclusionSerum lipase on POD 1 outperformed serum amylase on POD 1 in predicting CR-POPF after PD. The combination of POD 1 serum lipase and POD 3 serum CRP provides a reliable predicting model for CR-POPF.