There is little research that focuses on the relationship between the gut, metabolism, nutritional support and COVID-19. As a group of Chinese physicians, nutritionists and scientists working on the frontline treating COVID-19 patients, we aim to integrate our experiences and the current clinical evidence to address this pressing issue in this article. Based on our clinical observations and available evidence, we recommend the following practice. Firstly, the Nutritional Risk Screening 2002 tool should be used routinely and periodically; for patients with a score ≥3, oral nutritional supplements should be given immediately. Secondly, for patients receiving the antiviral agents lopinavir/ritonavir, gastrointestinal side effects should be monitored for and timely intervention provided. Thirdly, for feeding, the enteral route should be the first choice. In patients undergoing mechanical ventilation, establishing a jejunal route as early as possible can guarantee the feeding target being achieved if gastric dilatation occurs. Fourthly, we suggest a permissive underfeeding strategy for severe/critical patients admitted to the intensive care unit during the first week of admission, with the energy target no more than 20 kcal/kg/day (for those on mechanical ventilation, this target may be lowered to 10–15 kcal/kg/day) and the protein target around 1.0–1.2 g/kg/day. If the inflammatory condition is significantly alleviated, the energy target may be gradually increased to 25–30 kcal/kg/day and the protein target to 1.2–1.5 g/kg/day. Fifthly, supplemental parenteral nutrition should be used with caution. Lastly, omega-3 fatty acids may be used as immunoregulators, intravenous administration of omega-3 fatty emulsion (10 g/day) at an early stage may help to reduce the inflammatory reaction.

Summary Studies for sepsis prediction using machine learning are developing rapidly in medical science recently. In this review, we propose a set of new evaluation criteria and reporting standards to assess 21 qualified machine learning models for quality analysis based on PRISMA. Our assessment shows that (1.) the definition of sepsis is not consistent among the studies; (2.) data sources and data preprocessing methods, machine learning models, feature engineering, and inclusion types vary widely among the studies; (3.) the closer to the onset of sepsis, the higher the value of AUROC is; (4.) the improvement in AUROC is primarily due to using machine learning as a feature engineering tool; (5.) deep neural networks coupled with Sepsis-3 diagnostic criteria tend to yield better results on the time series data collected from patients with sepsis. The new evaluation criteria and reporting standards will facilitate the development of improved machine learning models for clinical applications.

No abstract