Machine learning uncovers blood test patterns subphenotypes at hospital admission discerning increased 30-day ICU mortality rates in COVID-19 elderly patients

Introduction: Since March 2020, a number of SARS-CoV-2 patients have frequently required intensive care unit (ICU) admission, associated with moderate survival outcomes and an increasing economic burden. Elderly patients are among the most numerous, due to previous comorbidities and complications they develop during hospitalization [1]. For this reason, a reliable early risk stratification tool could help estimate an early prognosis and allow for an appropriate resources allocation in favour of the most vulnerable and critically ill patients. Method(s): This retrospective study includes data from two Spanish hospitals, HU12O (Madrid) and HCUV (Valencia), from 193 patients aged > 64 with COVID-19 between February and November 2020 who were admitted to the ICU. Variables include demographics, full-blood-count (FBC) tests and clinical outcomes. Machine learning applied a non-linear dimensionality reduction by t-distributed stochastic neighbor embedding (t-SNE) [2];then hierarchical clustering on the t-SNE output was performed. The number of clinically relevant subphenotypes was chosen by combining silhouette and elbow coefficients, and validated through exploratory analysis. Result(s): We identified five subphenotypes with heterogeneous interclustering age and FBC patterns (Fig. 1). Cluster 1 was the 'healthiest' phenotype, with 2% 30-day mortality and characterized by moderate leukocytes and eosinophils. Cluster 5, the severe phenotype, showed 44% 30-day mortality and was characterized by the highest leukocyte, neutrophil and platelet count and minimal monocytes and lymphocyte count. Clusters 2-4 displayed intermediate mortality rates (20-28%). Conclusion(s): The findings of this preliminary report of Eld-ICUCOV19 patients suggest the patient's FBC and age can display discriminative patterns associated with disparate 30-day ICU mortality rates.

Prediction of the Estimated Time of Arrival of container ships on short-sea shipping: A pragmatical analysis

Fighting against climate change and global warming is one of the biggest challenges faced by the Maritime Industry nowadays to make the supply chain greener and environmentally sustainable. Cutting greenhouse gases (GHG) emissions and decarbonizing the international shipping has been a paramount activity for the International Maritime Organization (IMO) since the first set of international mandatory measures to improve ships' energy efficiency and reduce CO2 emissions per transport work, as part of the International Convention for the Prevention of Pollution from Ships (MARPOL) released in 2011. Besides that, changes in consumption habits around the globe (i.e., digitalization and growth of e-commerce) plus disruptive events like the COVID-19 or the blocking of the Suez Canal, to name only a few, have also highlighted the need for building more resilient maritime transport networks. In this work, a pragmatical analysis of the principal machine learning algorithms has been carried out to provide a qualitative prediction of the Estimate Time of Arrival (ETA) of container vessels applied to short-sea shipping where the distance between ports is reduced. By exploiting both, the Automatic Identification System (AIS) and meteorological data gathered over a desired area of interest, the developed approach delivers a model capable of predicting the ETA of ships where the reaction time of the stakeholders involved in the management of the Port Call is very reduced (i.e., less than two hours of sailing between ports) and therefore, tolerance for error is low. Very positive results were obtained for the training dataset collected under real conditions for more than a year. The best results were obtained by the RF model with a Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) of 11.31 and 19.56 minutes respectively. © 2003-2012 IEEE.

Association of sedation level and 28-day mortality: a COVID-19 cohort multicentre study sub analysis

Introduction: COVID-19 started in Wuhan (China) in December 2019 [1]. World pandemic was declared by the WHO in March 2020 [2]. Since then, millions of patients have been infected worldwide. Our group published in March 2021 a multicentre study analysing the prevalence and risk factors for delirium in critically ill patients with COVID-19 infection [3]. In this sub analysis of the main study, the primary outcome was the association of sedation level since ICU admission with 28-day mortality in patients admitted to ICU due to COVID-19. Methods: The exposure tested against mortality was excessive sedation (in the coma range) defined as 'Did patients have a documented sedation score in the coma range at any point this day? RASS = - 4 or - 5;SAS = 2 or 1;MAAS = 0 or 1;Ramsay 5 or 6'. Data from 2017 patients was available for analysis, collected after patients' admission to ICU. Other covariates analyzed were baseline patient characteristics, medical history and treatment applied in the ICU. Logistic regression was used in all analyses and results presented as odd ratios (OR) with 95% confidence intervals. Results: Deep sedation (RASS = - 4 or - 5, SAS = 2 or 1, MAAS = 0 or 1, Ramsay 5 or 6) was positively and significantly associated to mortality within 28 days since ICU admission. P value was 0.012 and the OR was 2.00 with a 95% confidence interval of 1.16-3.45. Conclusions: As shown by this sub analysis, deep sedation increases mortality rates in critically ill COVID 19 patients. We should try to decrease sedation levels to avoid RASS of - 4 and - 5 to favor patients' outcomes admitted to the ICU.