scholarly journals Predicting 90-day survival of patients with COVID-19: Survival of Severely Ill COVID (SOSIC) scores

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Matthieu Schmidt ◽  
Bertrand Guidet ◽  
Alexandre Demoule ◽  
Maharajah Ponnaiah ◽  
Muriel Fartoukh ◽  
...  

Abstract Background Predicting outcomes of critically ill intensive care unit (ICU) patients with coronavirus-19 disease (COVID-19) is a major challenge to avoid futile, and prolonged ICU stays. Methods The objective was to develop predictive survival models for patients with COVID-19 after 1-to-2 weeks in ICU. Based on the COVID–ICU cohort, which prospectively collected characteristics, management, and outcomes of critically ill patients with COVID-19. Machine learning was used to develop dynamic, clinically useful models able to predict 90-day mortality using ICU data collected on day (D) 1, D7 or D14. Results Survival of Severely Ill COVID (SOSIC)-1, SOSIC-7, and SOSIC-14 scores were constructed with 4244, 2877, and 1349 patients, respectively, randomly assigned to development or test datasets. The three models selected 15 ICU-entry variables recorded on D1, D7, or D14. Cardiovascular, renal, and pulmonary functions on prediction D7 or D14 were among the most heavily weighted inputs for both models. For the test dataset, SOSIC-7’s area under the ROC curve was slightly higher (0.80 [0.74–0.86]) than those for SOSIC-1 (0.76 [0.71–0.81]) and SOSIC-14 (0.76 [0.68–0.83]). Similarly, SOSIC-1 and SOSIC-7 had excellent calibration curves, with similar Brier scores for the three models. Conclusion The SOSIC scores showed that entering 15 to 27 baseline and dynamic clinical parameters into an automatable XGBoost algorithm can potentially accurately predict the likely 90-day mortality post-ICU admission (sosic.shinyapps.io/shiny). Although external SOSIC-score validation is still needed, it is an additional tool to strengthen decisions about life-sustaining treatments and informing family members of likely prognosis.

2020 ◽  
Author(s):  
Sujeong Hur ◽  
Ji Young Min ◽  
Junsang Yoo ◽  
Kyunga Kim ◽  
Chi Ryang Chung ◽  
...  

BACKGROUND Patient safety in the intensive care unit (ICU) is one of the most critical issues, and unplanned extubation (UE) is considered as the most adverse event for patient safety. Prevention and early detection of such an event is an essential but difficult component of quality care. OBJECTIVE This study aimed to develop and validate prediction models for UE in ICU patients using machine learning. METHODS This study was conducted an academic tertiary hospital in Seoul. The hospital had approximately 2,000 inpatient beds and 120 intensive care unit (ICU) beds. The number of patients, on daily basis, was approximately 9,000 for the out-patient. The number of annual ICU admission was approximately 10,000. We conducted a retrospective study between January 1, 2010 and December 31, 2018. A total of 6,914 extubation cases were included. We developed an unplanned extubation prediction model using machine learning algorithms, which included random forest (RF), logistic regression (LR), artificial neural network (ANN), and support vector machine (SVM). For evaluating the model’s performance, we used area under the receiver operator characteristic curve (AUROC). Sensitivity, specificity, positive predictive value negative predictive value, and F1-score were also determined for each model. For performance evaluation, we also used calibration curve, the Brier score, and the Hosmer-Lemeshow goodness-of-fit statistic. RESULTS Among the 6,914 extubation cases, 248 underwent UE. In the UE group, there were more males than females, higher use of physical restraints, and fewer surgeries. The incidence of UE was more likely to occur during the night shift compared to the planned extubation group. The rate of reintubation within 24 hours and hospital mortality was higher in the UE group. The UE prediction algorithm was developed, and the AUROC for RF was 0.787, for LR was 0.762, for ANN was 0.762, and for SVM was 0.740. CONCLUSIONS We successfully developed and validated machine learning-based prediction models to predict UE in ICU patients using electronic health record data. The best AUROC was 0.787, which was obtained using RF. CLINICALTRIAL N/A


2012 ◽  
Vol 44 (18) ◽  
pp. 865-877 ◽  
Author(s):  
Sudhakar Aare ◽  
Peter Radell ◽  
Lars I. Eriksson ◽  
Yi-Wen Chen ◽  
Eric P. Hoffman ◽  
...  

Severe muscle wasting and loss of muscle function in critically ill mechanically ventilated intensive care unit (ICU) patients have significant negative consequences on their recovery and rehabilitation that persist long after their hospital discharge; moreover, the underlying mechanisms are unclear. Mechanical ventilation (MV) and immobilization-induced modifications play an important role in these consequences, including endotoxin-induced sepsis. The present study aims to investigate how sepsis aggravates ventilator and immobilization-related limb muscle dysfunction. Hence, biceps femoris muscle gene expression was investigated in pigs exposed to ICU intervention, i.e., immobilization, sedation, and MV, alone or in combination with sepsis, for 5 days. In previous studies, we have shown that ICU intervention alone or in combination with sepsis did not affect muscle fiber size on day 5, but a significant decrease was observed in single fiber maximal force normalized to cross-sectional area (specific force) when sepsis was added to the ICU intervention. According to microarray data, the addition of sepsis to the ICU intervention induced a deregulation of >500 genes, such as an increased expression of genes involved in chemokine activity, kinase activity, and transcriptional regulation. Genes involved in the regulation of the oxidative stress response and cytoskeletal/sarcomeric and heat shock proteins were on the other hand downregulated when sepsis was added to the ICU intervention. Thus, sepsis has a significant negative effect on muscle function in critically ill ICU patients, and chemokine activity and heat shock protein genes are forwarded to play an instrumental role in this specific muscle wasting condition.


2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Qiangrong Zhai ◽  
Zi Lin ◽  
Hongxia Ge ◽  
Yang Liang ◽  
Nan Li ◽  
...  

AbstractThe number of critically ill patients has increased globally along with the rise in emergency visits. Mortality prediction for critical patients is vital for emergency care, which affects the distribution of emergency resources. Traditional scoring systems are designed for all emergency patients using a classic mathematical method, but risk factors in critically ill patients have complex interactions, so traditional scoring cannot as readily apply to them. As an accurate model for predicting the mortality of emergency department critically ill patients is lacking, this study’s objective was to develop a scoring system using machine learning optimized for the unique case of critical patients in emergency departments. We conducted a retrospective cohort study in a tertiary medical center in Beijing, China. Patients over 16 years old were included if they were alive when they entered the emergency department intensive care unit system from February 2015 and December 2015. Mortality up to 7 days after admission into the emergency department was considered as the primary outcome, and 1624 cases were included to derive the models. Prospective factors included previous diseases, physiologic parameters, and laboratory results. Several machine learning tools were built for 7-day mortality using these factors, for which their predictive accuracy (sensitivity and specificity) was evaluated by area under the curve (AUC). The AUCs were 0.794, 0.840, 0.849 and 0.822 respectively, for the SVM, GBDT, XGBoost and logistic regression model. In comparison with the SAPS 3 model (AUC = 0.826), the discriminatory capability of the newer machine learning methods, XGBoost in particular, is demonstrated to be more reliable for predicting outcomes for emergency department intensive care unit patients.


2018 ◽  
Vol 46 (3) ◽  
pp. 297-303 ◽  
Author(s):  
S. J. Ebmeier ◽  
M. Barker ◽  
M. Bacon ◽  
R. C. Beasley ◽  
R. Bellomo ◽  
...  

The influence of variables that might affect the accuracy of pulse oximetry (SpO2) recordings in critically ill patients is not well established. We sought to describe the relationship between paired SpO2/SaO2 (oxygen saturation via arterial blood gas analysis) in adult intensive care unit (ICU) patients and to describe the diagnostic performance of SpO2 in detecting low SaO2 and PaO2. A paired SpO2/SaO2 measurement was obtained from 404 adults in ICU. Measurements were used to calculate bias, precision, and limits of agreement. Associations between bias and variables including vasopressor and inotrope use, capillary refill time, hand temperature, pulse pressure, body temperature, oximeter model, and skin colour were estimated. There was no overall statistically significant bias in paired SpO2/SaO2 measurements; observed limits of agreement were +/-4.4%. However, body temperature, oximeter model, and skin colour, were statistically significantly associated with the degree of bias. SpO2 <89% had a sensitivity of 3/7 (42.9%; 95% confidence intervals, CI, 9.9% to 81.6%) and a specificity of 344/384 (89.6%; 95% CI 86.1% to 92.5%) for detecting SaO2 <89%. The absence of statistically significant bias in paired SpO2/SaO2 in adult ICU patients provides support for the use of pulse oximetry to titrate oxygen therapy. However, SpO2 recordings alone should be used cautiously when SaO2 recordings of 4.4% higher or lower than the observed SpO2 would be of concern. A range of variables relevant to the critically ill had little or no effect on bias.


2016 ◽  
Vol 54 (7) ◽  
pp. 1918-1921 ◽  
Author(s):  
Joerg Steinmann ◽  
Jan Buer ◽  
Peter-Michael Rath

We retrospectively analyzed the performance and relevance of the SeptiFast assay in detectingAspergillus fumigatusDNA in whole blood samples from 38 critically ill intensive care unit (ICU) patients with probable or proven invasive aspergillosis (IA) and 100 ICU patients without IA. The assay exhibited 66% sensitivity, 98% specificity, a 93% positive predictive value, and an 88% negative predictive value.A. fumigatusDNAemia was associated with poor outcome.


2013 ◽  
Vol 45 (8) ◽  
pp. 312-320 ◽  
Author(s):  
Sudhakar Aare ◽  
Peter Radell ◽  
Lars I. Eriksson ◽  
Hazem Akkad ◽  
Yi-Wen Chen ◽  
...  

Severe muscle wasting is a debilitating condition in critically ill intensive care unit (ICU) patients, characterized by general muscle weakness and dysfunction, resulting in a prolonged mobilization, delayed weaning from the ventilator, and a decreased quality of life post-ICU. The mechanisms underlying limb muscle weakness in ICU patients are complex and involve the impact of primary disease, but also factors common to critically ill ICU patients such as sepsis, mechanical ventilation (MV), immobilization, and systemic administration of corticosteroids (CS). These factors may have additive negative effects on skeletal muscle structure and function, but their respective role alone remain unknown. The primary aim of this study was to examine how CS administration potentiates ventilator and immobilization-related limb muscle dysfunction at the gene level. Comparing biceps femoris gene expression in pigs exposed to MV and CS for 5 days with only MV pigs for the same duration of time showed a distinct deregulation of 186 genes according to microarray. Surprisingly, the decreased force-generation capacity at the single muscle fiber reported in response to the addition of CS administration in mechanically ventilated and immobilized pigs was not associated with an additional upregulation of proteolytic pathways. On the other hand, an altered expression of genes regulating kinase activity, cell cycle, transcription, channel regulation, oxidative stress response, cytoskeletal, sarcomeric, and heat shock protein, as well as protein synthesis at the translational level, appears to play an additive deleterious role for the limb muscle weakness in immobilized ICU patients.


2007 ◽  
Vol 136 (8) ◽  
pp. 1009-1019 ◽  
Author(s):  
M. E. FALAGAS ◽  
E. A. KARVELI ◽  
I. I. SIEMPOS ◽  
K. Z. VARDAKAS

SUMMARYThere has been increasing concern regarding the rise ofAcinetobacterinfections in critically ill patients. We extracted information regarding the relative frequency ofAcinetobacterpneumonia and bacteraemia in intensive-care-unit (ICU) patients and the antimicrobial resistance ofAcinetobacterisolates from studies identified in electronic databases.Acinetobacterinfections most frequently involve the respiratory tract of intubated patients andAcinetobacterpneumonia has been more common in critically ill patients in Asian (range 4–44%) and European (0–35%) hospitals than in United States hospitals (6–11%). There is also a gradient in Europe regarding the proportion of ICU-acquired pneumonias caused byAcinetobacterwith low numbers in Scandinavia, and gradually rising in Central and Southern Europe. A higher proportion ofAcinetobacterisolates were resistant to aminoglycosides and piperacillin/tazobactam in Asian and European countries than in the United States. The data suggest thatAcinetobacterinfections are a growing threat affecting a considerable proportion of critically ill patients, especially in Asia and Europe.


2021 ◽  
Vol 5 (Supplement_2) ◽  
pp. 850-850
Author(s):  
Shih-Ching Lo ◽  
Yu-Chin Hsiao ◽  
Ying-Ru Chen ◽  
Hsing-Chun Lin

Abstract Objectives Aggressive nutritional intervention may improve the outcomes of critically ill patients. Therefore, the National Health Insurance Administration (NHIA) in Taiwan revised its relevant fee schedule. On October 1, 2019, nutritional care items for intensive care unit (ICU) patients, covered by the NHIA under the category of nutritional care fees, were introduced to reflect real clinical needs. Methods This retrospective cohort study was conducted in a medical center ICU. The study period was January 1, 2019 to May 31, 2020, before and after the start of national health insurance (NHI) coverage of new nutritional care items for ICU patients. A total of 5292 patients were recruited and divided into two groups based on timing of NHI coverage. There were 1591 patients included in the analysis (751 in the non-NHI group and 840 in the NHI group). In the NHI group, the following nutritional protocol was implemented: First visit was at 48hr following admission to the ICU with 2 follow up visits over the next 5 days, then 3 visits the following week. Patient demographics, daily nutritional data, and outcomes were collected to investigate the impact of this protocol. Results Both groups were given the same nutritional intervention initially. However, there were significant differences in nutritional intervention following the incorporation of this treatment protocol in the ICU. Closely monitored nutritional intervention met critical requirements without overfeeding and led to shorter ICU stays (non-NHI 8.11 ± 6.69 days vs NHI 7.12 ± 7.43 days, p &lt; 0.01). Conclusions Nutritional care plan based on frequent assessments and interventions by dietitians is associated with reduced ICU stays for critically ill patients. Funding Sources None.


2021 ◽  
Vol 12 ◽  
Author(s):  
François Mallet ◽  
Léa Diouf ◽  
Boris Meunier ◽  
Magali Perret ◽  
Frédéric Reynier ◽  
...  

IntroductionWe analysed blood DNAemia of TTV and four herpesviruses (CMV, EBV, HHV6, and HSV-1) in the REAnimation Low Immune Status Marker (REALISM) cohort of critically ill patients who had presented with either sepsis, burns, severe trauma, or major surgery. The aim was to identify common features related to virus and injury-associated pathologies and specific features linking one or several viruses to a particular pathological context.MethodsOverall and individual viral DNAemia were measured over a month using quantitative PCR assays from the 377 patients in the REALISM cohort. These patients were characterised by clinical outcomes [severity scores, mortality, Intensive Care Unit (ICU)-acquired infection (IAI)] and 48 parameters defining their host response after injury (cell populations, immune functional assays, and biomarkers). Association between viraemic event and clinical outcomes or immune markers was assessed using χ2-test or exact Fisher’s test for qualitative variables and Wilcoxon test for continuous variables.ResultsThe cumulative incidence of viral DNAemia increased from below 4% at ICU admission to 35% for each herpesvirus during the first month. EBV, HSV1, HHV6, and CMV were detected in 18%, 12%, 10%, and 9% of patients, respectively. The incidence of high TTV viraemia (&gt;10,000 copies/ml) increased from 11% to 15% during the same period. Herpesvirus viraemia was associated with severity at admission; CMV and HHV6 viraemia correlated with mortality during the first week and over the month. The presence of individual herpesvirus during the first month was significantly associated (p &lt; 0.001) with the occurrence of IAI, whilst herpesvirus DNAemia coupled with high TTV viraemia during the very first week was associated with IAI. Herpesvirus viraemia was associated with a lasting exacerbated host immune response, with concurrent profound immune suppression and hyper inflammation, and delayed return to immune homeostasis. The percentage of patients presenting with herpesvirus DNAemia was significantly higher in sepsis than in all other groups. Primary infection in the hospital and high IL10 levels might favour EBV and CMV reactivation.ConclusionIn this cohort of ICU patients, phenotypic differences were observed between TTV and herpesviruses DNAemia. The higher prevalence of herpesvirus DNAemia in sepsis hints at further studies that may enable a better in vivo understanding of host determinants of herpesvirus viral reactivation. Furthermore, our data suggest that EBV and TTV may be useful as additional markers to predict clinical deterioration in ICU patients.


2018 ◽  
Vol 33 (8) ◽  
pp. 546-553 ◽  
Author(s):  
Iván Sánchez Fernández ◽  
Arnold J. Sansevere ◽  
Marina Gaínza-Lein ◽  
Kush Kapur ◽  
Tobias Loddenkemper

The aim of this study was to evaluate the performance of models predicting in-hospital mortality in critically ill children undergoing continuous electroencephalography (cEEG) in the intensive care unit (ICU). We evaluated the performance of machine learning algorithms for predicting mortality in a database of 414 critically ill children undergoing cEEG in the ICU. The area under the receiver operating characteristic curve (AUC) in the test subset was highest for stepwise selection/elimination models (AUC = 0.82) followed by least absolute shrinkage and selection operator (LASSO) and support vector machine with linear kernel (AUC = 0.79), and random forest (AUC = 0.71). The explanatory models had the poorest discriminative performance (AUC = 0.63 for the model without considering etiology and AUC = 0.45 for the model considering etiology). Using few variables and a relatively small number of patients, machine learning techniques added information to explanatory models for prediction of in-hospital mortality.


Sign in / Sign up

Export Citation Format

Share Document