Early prediction of sepsis in intensive care units: A comparative analysis based on optimization techniques and committees

The early prediction of sepsis in intensive care units using clinical data is the objective of the PhysioNet/Computing in Cardiology Challenge 2019. In this paper, a machine learning approach is presented which uses an optimized Random Forest for prediction of a septic condition. After an initial data augmentation step, a customized learning process is performed for the trees to consider imbalance in the dataset. Finally, a feature reduction is implemented and the forest is trimmed to 50 trees for an optimal classification in terms of run time and accuracy. Using a 10-fold cross-validation on the complete training dataset, a mean utility score of 0.376 is achieved. In the final submission, a normalized observed utility score of 0.296 on the full test set is achieved. Our team name is The Septic Think Tank (final rank: 21).

Severe sepsis is associated with a high mortality rate, but the detailed epidemiology of sepsis is not well known in Polish hospitals. The aim of the study was to determine the prevalence and incidence of severe sepsis in Polish intensive care units (ICUs). Two one-day, point-prevalence studies were performed on March 8th, 2012 and March 13th, 2013. An online questionnaire was sent to 320 accredited ICUs. Demographic data regarding hospitals, ICUs, number of patients with severe sepsis and septic shock, and number of patients mechanically ventilated with a central catheter or a urinary catheter were collected. The one-day prevalence of severe sepsis in ICUs was calculated, and the annual incidence of severe sepsis in Poland was estimated from the prevalence rate and the mean length of stay in ICUs. 1398 patients participated in the study in 2012, which accounted for 50% of all ICU beds registered by the National Health Care (NHC) system; 860 patients participated in 2013 (30% of all ICU beds). The daily prevalence of severe sepsis in ICUs was 26% in 2012 and 22% in 2013. Based on the data provided by the NHC system, the number of severe sepsis patients treated in accredited ICUs in Poland amounted to 24,905 patients per year, and the incidence of severe sepsis was 65/100,000 cases per year. Severe sepsis was observed in one-fourth of patients treated in ICUs in Poland. However, the actual number of severe sepsis patients is at least 2 times higher because many patients with severe sepsis were treated outside accredited ICUs. Severe sepsis constitutes a major health problem in Poland.

To investigate the power of the simplified acute physiology score III (SAPSIII) for prediction of outcome for patients with severe sepsis admitted to the intensive care unit (ICU). A retrospective study was conducted. 677 severe sepsis patients with age≥18 years old and the survival time in emergency ICU≥24 hours admitted to the emergency ICU of Beijing Chaoyang Hospital Affiliated to Capital Medical University from January 2008 to December 2011 were enrolled. The acute physiology and chronic health evaluation II (APACHEII), sequential organ failure assessment (SOFA), SAPSII, SAPSIII, and mortality in emergency department sepsis (MEDS) scores were recorded based on the poorest value within 24 hours of ICU admission. The 28-day result as denoted as survival or death was considered as the end point of the study. The ability to predict mortality by the score systems was assessed by using receiver operating characteristic (ROC) curve analysis and binary logistic regression models. Among the 677 patients with severe sepsis, 284 cases died within 28 days after admission, and the mortality rate was 41.9%. Compared with survivors, the patients in non-survival group was older with higher APACHEII, SOFA, SAPSII, SAPSIII, and MEDS scores and higher ratio of underlying diseases, such as primary hypertension and renal dysfunction, and they had more organ injury, higher ratio of lung infection and bacterial infection (P<0.05 or P<0.01). It was identified by logistic regression that the APACHEII, SOFA, SAPSII, SAPSIII and MEDS scores were significantly independent factors in 28-day death prediction in patients with severe sepsis (all P=0.000). The rank of areas under the ROC curve (AUC) from high to low were MEDS (0.970), APACHEII (0.893), SAPSIII (0.875), SOFA (0.871), and SAPSII (0.860), respectively. SAPSIII score and APACHEII, SOFA, SAPSII scores were found to have an equivalent capacity in predicting the prognosis (all P>0.05). The MEDS score in predicting the prognosis was obviously better than that of APACHEII, SOFA, SAPSII, and SAPSIII scores (all P<0.05 ). The MEDS score showed the best sensitivity (91.5%), and specificity (89.1%). The 28-day mortality in cases of MEDS≥11 was 85.8%. (1) For patients with severe sepsis who were admitted to ICU, MEDS was superior to APACHEII, SOFA, SAPSII, and SAPSIII scores in predicting prognosis. MEDS≥11 may indicate a higher mortality rate. (2) SAPSIII score has comparable predictive capability with APACHEII, SOFA and SAPSII scores may be recommended for prediction of the prognosis of patients with severe sepsis in ICU. But the SAPSIII score is unsuitable for predicting the prognosis of patients with acute sepsis in ICU options, and it is not superior to that of SAPSIII score in predicting prognosis of patients with sepsis in the emergency ICU than other score systems.

The simple scoring systems for predicting the outcome of sepsis in intensive care units (ICUs) are few, especially for limited-resource settings. Therefore, this study aimed to evaluate the accuracy of the quick Sequential (Sepsis-Related) Organ Failure Assessment (qSOFA) score in predicting the mortality of ICU patients with sepsis in Vietnam. We did a multicenter cross-sectional study of patients with sepsis (≥18 years old) presenting to 15 adult ICUs throughout Vietnam on the specified days (i.e., 9th January, 3rd April, 3rd July, and 9th October) representing the different seasons of 2019. The primary and secondary outcomes were the hospital and ICU all-cause mortalities, respectively. The area under the receiver operating characteristic curve (AUROC) was calculated to determine the discriminatory ability of the qSOFA score for deaths in the hospital and ICU. The cut-off value of the qSOFA scores was determined by the receiver operating characteristic curve analysis. Upon ICU admission, factors associated with the hospital and ICU mortalities were assessed in univariable and multivariable logistic models. Of 252 patients, 40.1% died in the hospital, and 33.3% died in the ICU. The qSOFA score had a poor discriminatory ability for both the hospital (AUROC: 0.610 [95% CI: 0.538 to 0.681]; cut-off value: ≥2.5; sensitivity: 34.7%; specificity: 84.1%; PAUROC = 0.003) and ICU (AUROC: 0.619 [95% CI: 0.544 to 0.694]; cutoff value: ≥2.5; sensitivity: 36.9%; specificity: 83.3%; PAUROC = 0.002) mortalities. However, multivariable logistic regression analyses show that the qSOFA score of 3 was independently associated with the increased risk of deaths in both the hospital (adjusted odds ratio, AOR: 3.358; 95% confidence interval, CI: 1.756 to 6.422) and the ICU (AOR: 3.060; 95% CI: 1.651 to 5.671). In our study, despite having a poor discriminatory value, the qSOFA score seems worthwhile in predicting mortality in ICU patients with sepsis in limited-resource settings. Clinical trials registry-India: CTRI/2019/01/016898.

BackgroundThe simple scoring systems for predicting the outcome of sepsis in intensive care units (ICUs) are few, especially for limited-resource settings. Therefore, this study aimed to evaluate the accuracy of the quick Sequential (Sepsis-Related) Organ Failure Assessment (qSOFA) score in predicting the mortality of ICU patients with sepsis in Vietnam.MethodsWe did a multicenter cross-sectional study of patients with sepsis (≥18 years old) presenting to 15 adult ICUs throughout Vietnam on the specified days (i.e., 9th January, 3rd April, 3rd July, and 9th October) representing the different seasons of 2019. The primary and secondary outcomes were the hospital and ICU all-cause mortalities, respectively. The area under the receiver operating characteristic curve (AUROC) was calculated to determine the discriminatory ability of the qSOFA score for deaths in the hospital and ICU. The cut-off value of the qSOFA scores was determined by the receiver operating characteristic curve analysis. Upon ICU admission, factors associated with the hospital and ICU mortalities were assessed in univariable and multivariable logistic models.ResultsOf 252 patients, 40.1% died in the hospital, and 33.3% died in the ICU. The qSOFA score had a poor discriminatory ability for both the hospital (AUROC: 0.610 [95% CI: 0.538 to 0.681]; cut-off value: ≥2.5; sensitivity: 34.7%; specificity: 84.1%; PAUROC = 0.003) and ICU (AUROC: 0.619 [95% CI: 0.544 to 0.694]; cutoff value: ≥2.5; sensitivity: 36.9%; specificity: 83.3%; PAUROC = 0.002) mortalities. However, multivariable logistic regression analyses show that the qSOFA score of 3 was independently associated with the increased risk of deaths in both the hospital (adjusted odds ratio, AOR: 3.358; 95% confidence interval, CI: 1.756 to 6.422) and the ICU (AOR: 3.060; 95% CI: 1.651 to 5.671).ConclusionIn our study, despite having a poor discriminatory value, the qSOFA score seems worthwhile in predicting mortality in ICU patients with sepsis in limited-resource settings.Clinical trial registrationClinical trials registry–India: CTRI/2019/01/016898

Objective To study and discuss the clinical effect of continuous renal replacement therapy (CRRT) in the treatment of patients with sepsis in intensive care unit (ICU). Methods From January 2014 to February 2017, 100 patients with sepsis in the ICU of Yuncheng Central Hospital were selected and randomly divided into control group and observation group, with 50 cases in each group.The control group was treated with hemoperfusion treatment, the observation group was treated with CRRT.The blood coagulation function, clinical efficacy, renal function, inflammatory factors index were compared between the two groups. Results The total effective rate of the observation group was 96%, which was significantly higher than 80% of the control group (χ2=6.061, P<0.05). After treatment, the blood creatinine, urea nitrogen in the observation group were (94.09 ±20.69)μmol/L, (8.94±2.87)mmol/L, respectively, which were lower than those in the control group [(119.43±26.57)μmol/L, (12.37±3.70)mmol/L] (t=5.321, 5.180, all P<0.05). The APTT, PT of the observation group were (19.10±2.14)s, (8.24±0.97)s, respectively, which were shorter than those of the control group [(21.84±2.75)s, (9.32±1.15)s] (t=5.560, 5.076, all P<0.05). The levels of CRP, IL-6 of the observation group were (8.32±1.89)mg/L, (109.53±36.29)ng/L, respectively, which were lower than those of the control group [(12.65±3.47)mg/L, (148.36±43.64)ng/L] (t=7.749, 4.838, all P<0.05). Conclusion CRRT treatment for the sepsis patients during ICU monitoring can effectively improve the clinical curative effect, improve renal function, reduce the damage of blood coagulation function, inhibit the inflammatory reaction in the body, and is beneficial for the prognosis. Key words: Sepsis; ICU; Continuous renal replacement therapy; Kidney function tests; Blood coaguation factors; Chemokines

BackgroundSepsis, a severe systemic response to infection, frequently results in adverse outcomes, underscoring the urgency for prompt and accurate prognostic tools. Machine learning methods such as logistic regression, random forests, and CatBoost, have shown potential in early sepsis prediction. The study aimed to create and verify a machine learning model capable of early prognostic identification of patients with sepsis in intensive care units (ICUs).MethodsPatients adhering to inclusion and exclusion criteria from the MIMIC-IV v2.2 database were divided into a training set and a validation set in a 7:3 ratio. Initially, we employed difference analysis to assess the significance of each variable and subsequently screened relevant features with multinomial logistic regression analysis. Logistic regression, random forest, and CatBoost algorithms were used to construct machine learning models to predict rapid recovery, chronic critical illness, and mortality in sepsis. The models were compared through several evaluation indexes including precision, accuracy, recall, F1 score, and the area under the receiver-operating-characteristic curve(AUC) in the validation set to select the optimal model. The best model was visualized and interpreted utilizing the Shapley Additive explanations method.Results13174 sepsis patients were included. Post the screening process,26 clinical features were obtained to develop three distinct machine learning models. CatBoost exhibited superior performance among the three models with a weighted AUC of 0.771. The prognosis with the highest predictive performance was mortality (AUC = 0.804), followed by the prognoses of rapid recovery (AUC = 0.773) and chronic critical illness(AUC = 0.737). Urine output, respiratory rate, and temperature were the top three important features for the whole model prediction.ConclusionThe machine learning model developed leveraging the CatBoost algorithm demonstrates the latent capacity to identify sepsis prognosis early. It also suggests that interventions targeting factors such as urine output, respiratory status, and temperature in the early stage may potentially alter the adverse prognosis of sepsis patients. However, the model will still require further external validation in the future.

Background: Due to a scarcity of national epidemiologic data on sepsis, we performed a national cross-sectional survey to determine the epidemiologic characteristics of patients with sepsis in intensive care units (ICUs) in China. Methods: A national survey was performed in 44 hospitals which were selected by randomization according to the population from each province in mainland China. All septic patients in participating ICUs from December 1, 2015 to January 31, 2016 were included in the analysis. We recorded demographic, physiological and bacteriological data and followed up for 90 days or until death after inclusion. The incidence of and mortality due to sepsis in the ICU was calculated. The relationship between gross domestic product (GDP) and incidence of sepsis was also determined. Findings: Of 2,322 patients with sepsis included in the analysis, 786 (33.85%) patients had hospital-acquired sepsis with the most common infection site being the lung. The incidence of sepsis in the ICU was 20.6[95%CI 15.83-25.37] cases per 100 ICU admissions were diagnosed as sepsis. There was significant variation in the incidence of sepsis in the ICUs of different geographic regions. Ninety-day mortality due to sepsis was 35.49%, which yielded about 81,876 deaths in patients with ICU-treated sepsis every year in the mainland China. There was a significant relationship between the incidence of sepsis in the ICU and GDP per capita (R2=0.304, p=0.001) and the ratio of ICU beds to hospital beds (R2=0.270, p=0.003). Interpretation: Sepsis is a frequently occurring, life-threatening syndrome in the ICU with a high mortality in the mainland of China. Clinical Trial Number: This study was also registered in ClinicalTrials.gov (https://clinicaltrials.gov/show/NCT02448472). Funding Statement: This study was supported by a grant from the Development Center for Medical Science and Technology National Health and Family Planning Commission of the People’s Republic of China (WH2015-01-01), a grant from Jiangsu Provincial Key Medical Discipline (grant number: ZDXKA2016025) and a grant from Jiangsu Provincial Medical Talent (grant number: ZDRCA2016082). Declaration of Interests: The authors declare no competing interests. Ethics Approval Statement: The protocol of this study was approved by the ethics committees of Zhongda Hospital, School of Medicine, Southeast University (No. 2015ZDSYLL044.0), and every enrolled hospital. Patients admitted into the ICU were enrolled in this study following informed consent from the patient or their guardian.

Neutrophil CD64, C-reactive protein, and procalcitonin in the identification of sepsis in the ICU — Post-test probabilities

IntroductionPredictive scoring system are measures of disease severity used to predict outcomes including mortality. Such measurements compare the quality of patient care across ICUs. Several scoring system are available to predict sepsis outcome in the Emergency and ICU, two of which most recently used and upgraded are SOFA score (GOLD STANDARD), APACHE 4 and SAPS 3. Scores serve the purposes of assessing therapies, quality control, quality assurance and economic evaluation in intensive care. They allow timely identification of high-risk populations that require aggressive management and intervention. They also help in making objective prognoses and recommendations for clinicians as well as patients and their families. POSMI score was developed specifically to predict the severity of illness in patients with sepsis. It doesn't require software system for calculation and can be calculated in bedside. The POSMI score was derived from retrospective cohort of patients in USA and CHINA. To our knowledge it has never been validated in a prospective study in Indian patients.ObjectivesPrimary Objective: To validate the efficacy of POSMI SCORE in predicting mortality of sepsis in a tertiary care ICU of eastern India and to compare with SOFA score, SAPS 3 and APACHE 4 in doing so.MaterialsSTUDY AREA –70 bedded Intensive care unit (ICU) of tertiary care hospital in Eastern India. STUDY POPULATION –Convenience sample of 80 (patients admitted with sepsis in ICU. STUDY DURATION- 1ST JANUARY 2023- 30TH NOVEMBER 2023 STUDY DESIGN-prospective longitudinal observational study.Methodspatients with sepsis were evaluated by the above mentioned scores after 24hour of hospital admission. We followed the patients till hospital discharge or death.ResultsOur study included 80 patients whose mean age is (68.38 + 12.14 years) comprising of 63% male and 37% female population. Values of variables expressed as median in interquartile range are APACHE 4 - 76.5(27), SAPS 3 - 70 (13), SOFA -6.50(4) and POSMI -7.25(4). The association between scores and mortality show higher co relation (Spearman Rank) coefficient with POSMI (0.605) followed by SOFA (0.420), APACHE 4(0.386) and SAPS 3(0.285). The AUC (AREA UNDER CURVE) for POSMI score {0.903, 95% CONFIDENCE INTERVAL (CI) 0.837-0.968} is higher than SOFA score {0.778, 95% (CI) 0.668-0.888}, APACHE 4 {0.757, 95% (CI) 0.646-0.868} and SAPS 3 {0.690 95% (CI) 0.568-0.811}. This indicated that POSMI score has better discrimination than the other scores compared here.DiscussionPOSMI score can be used as an effective bedside tool for predicting mortality in patients with sepsis.

<b>Background:</b> With a rapidly aging population, there is increasing number of elderly and very elderly patients admitted to intensive care unit (ICU) for sepsis. <b>Aim:</b> To evaluate the clinical characteristics, short- and long-term outcomes of these patients. <b>Method:</b> This was a retrospective cohort study of ICU patients admitted for sepsis between July 2015 and October 2016. Patients were divided into young (&lt;65 years), elderly (65 to 79 years) and very elderly (≥80 years). Hospital and one-year mortality, ICU and hospital length of stay and functional status at hospital discharge were examined. <b>Results:</b> There were 668 patients. 37.3% were young, 43.1% were elderly and 19.6% were very elderly. Elderly and very elderly had more comorbidities. There was no difference in APACHE IIM (modified to exclude age points) (p=0.16) and SOFA score (p=0.54). Hospital mortality increased with age. Mortality was higher in very elderly (55.7%) versus young (39.4%) (p&lt;0.01), and elderly (48.6%) versus young (p=0.03), but no difference was seen between very elderly and elderly (p=0.18). Similar findings were observed in one-year mortality. There was no difference in median ICU length of stay (5 days in all groups), but median hospital length of stay was significantly longer in elderly (20 days) compared to young (14 days) (p=0.01). There were 357 hospital survivors. Functional status at hospital discharge was worse with age, with lower proportion of independence in activities of daily living in very elderly (31.0%) and elderly (52.0%) compared to young (70.9%) (p&lt;0.01 and p&lt;0.01 respectively). <b>Conclusion:</b> Elderly and very elderly patients admitted to ICU for sepsis had considerably poorer outcomes with a higher burden of care.

Objective To investigate the effect of 3 pediatric scores of critical illness including Pediatric Critical Illness Score (PCIS), Pediatric Logistic Organ Dysfunction Score 2 (PELOD-2), and Pediatric Multiple Organ Dysfunction Score (P-MODS) in estimating the prognosis of illness in children with sepsis in pediatric intensive care unit (PICU). Methods The data of 516 pediatric patients diagnosed as sepsis in PICU of Affiliated Hospital of Guangdong Medical University from June 2016 to June 2018 were retrospectively analyzed, and they were divided into survival group and death group according to the clinical outcome on the 28th day after admission.Then, the receiver operating characteristic (ROC) curve was plotted, and the efficiency of PCIS, PELOD-2 and P-MODS for predicting death was evaluated by the area under ROC curve (AUC). Results There were 488 pediatric patients survived, while 28 cases died during hospitalization.Compared with the survival group, the death group had a significantly lower PCIS score [86(82, 88) scores vs.89(84, 92)scores], and significantly higher PELOD-2 and P-MODS scores[PELOD-2: 6.5(4.0, 8.0) scores vs.0 (0, 2.0) scores, P-MODS: 3(2, 6) scores vs.1(1, 2) scores], and the differences were significant(Z=3 259.500, 14.228, 4.688, all P<0.05). ROC curve analysis showed that the AUCs of PCIS, PELOD-2 and P-MODS for predicting prognosis of pediatric patients with sepsis in PICU were 0.761, 0.916 and 0.761, respectively(Z=6.127, 14.228, 4.688, all P<0.05). Conclusions PCIS, PELOD-2 and P-MODS are effective and have good ability to assess the prognosis of pediatric patients with sepsis in the PICU.It seems that PELOD-2 is the most effective. Key words: Pediatric scores of critical illness; Pediatric intensive care unit; Sepsis; Prognostic evaluation

Studies on head injury-induced pituitary dysfunction are limited in number and conflicting results have been reported.To further clarify this issue, 29 consecutive patients (24 males), with severe (n = 21) or moderate (n = 8) head trauma, having a mean age of 37 ± 17 years were investigated in the immediate post-trauma period.All patients required mechanical ventilatory support for 8-55 days and were enrolled in the study within a few days before ICU discharge.Basal hormonal assessment included measurement of cortisol, corticotropin, free thyroxine (fT4), thyrotropin (TSH), testosterone (T) in men, estradiol (E2) in women, prolactin (PRL), and growth hormone (GH).Cortisol and GH levels were measured also after stimulation with 100 µg human corticotropin releasing hormone (hCRH) and 100 µg growth hormone releasing hormone (GHRH), respectively.Cortisol hyporesponsiveness was considered when peak cortisol concentration was less than 20 µg/dl following hCRH.TSH deficiency was diagnosed when a subnormal serum fT4 level was associated with a normal or low TSH.Hypogonadism was considered when T (males) or E2 (women) were below the local reference ranges, in the presence of normal PRL levels.Severe or partial GH deficiencies were defined as a peak GH below 3 µg/l or between 3 and 5 µg/l, respectively, after stimulation with GHRH.Twenty-one subnormal responses were found in 15 of the 29 patients (52%) tested; seven (24%) had hypogonadism, seven (24%) had cortisol hyporesponsiveness, five (17%) had hypothyroidism, and two patients (7%) had partial GH deficiency.These preliminary results suggest that a certain degree of hypopituitarism occurs in more than 50% of patients with moderate or severe head injury in the immediate post-trauma period, with cortisol hyporesponsiveness and hypogonadism being most common.Further studies are required to elucidate the pathogenesis of these abnormalities and to investigate whether they affect long-term morbidity.

Objective: to describe the role of nurses in preventing sepsis in adult intensive care units (ICU). Methodology: integrative literature review study. It was used the databases LILACS, BDENF, SCIELO, REDIB, BVS; through the CEDs: nursing, intensive care units, sepsis, prevention, in Portuguese and English. Inclusion criteria used: document type, articles, selected databases, published between the years 2016 and 2022. Results: the sample included 03 national articles, since they were written by nurse authors. Conclusion: It is considered that the research enabled the analysis of scientific productions about the knowledge, practices and strategies employed in the nurse’s performance for prevention of sepsis in ICU. With the purpose of obtaining a holistic view of the role of nurses facing patients who require intensive care, through a look at the completeness of the individual as a human being.

BackgroundWeaning from mechanical ventilation is one of the most common challenges in the intensive care unit (ICU). Most of predictive indices of weaning from mechanical ventilation are often inaccurate. This study was performed to assess the accuracy of diaphragmatic ultrasonography for predicting weaning outcome in mechanically ventilated patients with sepsis in ICU.ResultsSixty patients with sepsis in medical ICU were prospectively enrolled. All patients were ventilated in pressure support. Patients underwent a spontaneous breathing trial (SBT) on T-piece when they met all the following criteria: FiO2 < 0.6, PEEP ≤5 cmH2O, PaO2/FiO2 > 200, respiratory rate <30 breaths per minute, absence of fever, alert and cooperative, hemodynamic stability without or with low-dose vasoactive therapy support, and rapid shallow breathing index (RSBI)<105. During the trial, the patient was instructed to perform deep breathing to total lung capacity (TLC) and then exhaling to residual volume (RV) and the diaphragm was visualized in the 8th or 9th intercostal space between anterior and mid-axillary lines using a 3–5-MHz curved ultrasound probe to measure diaphragmatic excursion (DE) and a 7–11-MHz linear ultrasound probe to measure diaphragmatic thickness (DT) at TLC and RV, and the diaphragmatic thickness fraction (DTF) was calculated as percentage from the following formula (thickness at end inspiration—thickness at end expiration)/thickness at end expiration. According to weaning outcome, patients were divided into 2 groups: successful weaning group and weaning failure group. Weaning failure was defined as the inability to maintain spontaneous breathing for at least 48 h, without any form of ventilatory support.In the present study, right DTF of more than 37% and DE during deep breathing of more than 6.1 and 5.4 cm on the Rt and Lt side, respectively, were associated with successful weaning from MV. In the study, the sensitivities for right and left DE and DTF were 58.33, 62.5, and 58.33%, respectively, and the pooled specificities were 83.33, 83.33, and 100%, respectively, with p value = 0.032, 0.028, and 0.001, respectively. The area under curve (AUC) for Rt, Lt DE, and DTF were 0.701, 0.712, and 0.840, respectively. The present data indicate a satisfactory diagnostic accuracy in predicting extubation outcome.ConclusionsUltrasonography-based determination of diaphragm function by assessing DTF and DE can be used as predictor of weaning outcome in mechanically ventilated patients with sepsis.