scholarly journals Machine Learning Approaches to Predict In-Hospital Mortality among Neonates with Clinically Suspected Sepsis in the Neonatal Intensive Care Unit

2021 ◽  
Vol 11 (8) ◽  
pp. 695
Author(s):  
Jen-Fu Hsu ◽  
Ying-Feng Chang ◽  
Hui-Jun Cheng ◽  
Chi Yang ◽  
Chun-Yuan Lin ◽  
...  
Keyword(s):  
Machine Learning ◽  
Intensive Care Unit ◽  
Intensive Care ◽  
Hospital Mortality ◽  
Neonatal Intensive Care Unit ◽  
Clinical Features ◽  
Neonatal Intensive Care ◽  
Learning Approaches ◽  
Suspected Sepsis ◽  
Test Sets

Background: preterm and critically ill neonates often experience clinically suspected sepsis during their prolonged hospitalization in the neonatal intensive care unit (NICU), which can be the initial sign of final adverse outcomes. Therefore, we aimed to utilize machine learning approaches to predict neonatal in-hospital mortality through data-driven learning. Methods: a total of 1095 neonates who experienced clinically suspected sepsis in a tertiary-level NICU in Taiwan between August 2017 and July 2020 were enrolled. Clinically suspected sepsis was defined based on clinical features and laboratory criteria and the administration of empiric antibiotics by clinicians. The variables used for analysis included patient demographics, clinical features, laboratory data, and medications. The machine learning methods used included deep neural network (DNN), k-nearest neighbors, support vector machine, random forest, and extreme gradient boost. The performance of these models was evaluated using the area under the receiver operating characteristic curve (AUC). Results: the final in-hospital mortality of this cohort was 8.2% (90 neonates died). A total of 765 (69.8%) and 330 (30.2%) patients were randomly assigned to the training and test sets, respectively. Regarding the efficacy of the single model that most accurately predicted the outcome, DNN exhibited the greatest AUC (0.923, 95% confidence interval [CI] 0.953–0.893) and the best accuracy (95.64%, 95% CI 96.76–94.52%), Cohen’s kappa coefficient value (0.74, 95% CI 0.79–0.69) and Matthews correlation coefficient value (0.75, 95% CI 0.80–0.70). The top three most influential variables in the DNN importance matrix plot were the requirement of ventilator support at the onset of suspected sepsis, the feeding conditions, and intravascular volume expansion. The model performance was indistinguishable between the training and test sets. Conclusions: the DNN model was successfully established to predict in-hospital mortality in neonates with clinically suspected sepsis, and the machine learning algorithm is applicable for clinicians to gain insights and have better communication with families in advance.

scholarly journals Choice of Antibiotics in Late Neonatal Sepsis in the Extremely Low Birth Weight Infant

2003 ◽  
Vol 14 (1) ◽  
pp. 28-31 ◽  
Author(s):  
Tara R Allen ◽  
Orlando P da Silva
Keyword(s):  
Intensive Care Unit ◽  
Intensive Care ◽  
Birth Weight ◽  
Neonatal Intensive Care Unit ◽  
Neonatal Sepsis ◽  
Neonatal Intensive Care ◽  
Bacterial Resistance ◽  
Neonatal Infection ◽  
Extremely Low Birth Weight ◽  
Suspected Sepsis

OBJECTIVE: To review the choice of antibiotics in treating suspected late neonatal sepsis in infants weighing 1000 g or less in a neonatal intensive care unit.METHODS: Retrospective review of medical records.RESULTS: Ninety-six infants weighing 1000 g or less were admitted to the neonatal intensive care unit during the study period. Sixty-two infants survived beyond four days of life and had at least one sepsis workup done to exclude late neonatal infection. Of the 62 study patients, 42 (68%) were started on ampicillin and netilmicin (A/N) and 20 (32%) were started on vancomycin and ceftizoxime (V/C) as the antibiotics of choice, pending culture results. Of the patients started on A/N, 17 of 42 had a positive blood culture compared with 11 of 20 on V/C (40% versus 55%, P=0.40). The mean (±SD) birth weight of infants started on A/N was 793±133 g compared with a mean of 728±153 g in the group that received V/C (P=0.09). Seven patients died in the A/N group compared with three in the V/C group (16.7% versus 15%, P=0.84). In addition to the sepsis episode studied, before they were discharged from hospital, 21 of 42 (50%) infants in the A/N group had further workups for suspected sepsis, compared with 16 of 20 (80%) (P=0.048) infants initially given V/C.CONCLUSIONS: Ampicillin and netilmicin is a safe antibiotic combination for neonates suspected of late sepsis. This, in turn, may be important in reducing vancomycin overuse and the potential for bacterial resistance to this antimicrobial agent.

Machine Learning to Support Hemodynamic Intervention in the Neonatal Intensive Care Unit

2020 ◽  
Vol 47 (3) ◽  
pp. 435-448
Author(s):  
David Van Laere ◽  
Marisse Meeus ◽  
Charlie Beirnaert ◽  
Victor Sonck ◽  
Kris Laukens ◽  
...  
Keyword(s):  
Machine Learning ◽  
Intensive Care Unit ◽  
Intensive Care ◽  
Neonatal Intensive Care Unit ◽  
Neonatal Intensive Care

Association of Resident Duty Hour Restrictions, Level of Trainee, and Number of Available Residents with Mortality in the Neonatal Intensive Care Unit

2018 ◽  
Vol 35 (09) ◽  
pp. 911-918 ◽  
Author(s):  
Karin Clement ◽  
Guy Lacroix ◽  
Sylvie Bélanger ◽  
Anne-Sophie Julien ◽  
Bruno Piedboeuf ◽  
...  
Keyword(s):  
Intensive Care Unit ◽  
Intensive Care ◽  
Hospital Mortality ◽  
Neonatal Intensive Care Unit ◽  
Neonatal Intensive Care ◽  
Early Mortality ◽  
Hours Worked ◽  
Pediatric Residents ◽  
Level 3 ◽  
Resident Duty

Objective This article assesses the effect of reducing consecutive hours worked by residents from 24 to 16 hours on yearly total hours worked per resident in the neonatal intensive care unit (NICU) and evaluates the association of resident duty hour reform, level of trainee, and the number of residents present at admission with mortality in the NICU. Study Design This is a 6-year retrospective cohort study including all pediatric residents working in a Level 3 NICU (N = 185) and infants admitted to the NICU (N = 8,159). Adjusted odds ratios (aOR) were estimated for mortality with respect to Epoch (2008–2011 [24-hour shifts] versus 2011–2014 [16-hour shifts]), level of trainee, and the number of residents present at admission. Results The reduction in maximum consecutive hours worked was associated with a significant reduction of the median yearly total hours worked per resident in the NICU (381 hour vs. 276 hour, p < 0.01). Early mortality rate was 1.2% (50/4,107) before the resident duty hour reform and 0.8% (33/4,052) after the reform (aOR, 0.57; 95% confidence interval [CI], 0.33–0.98). Neither level of trainee (aOR, 1.22; 95% CI, 0.71–2.10; junior vs. senior) nor the number of residents present at admission (aOR, 2.08; 95% CI, 0.43–10.02, 5–8 residents vs. 0–2 residents) were associated with early mortality. Resident duty hour reform was not associated with hospital mortality (aOR, 0.73; 95% CI, 0.50–1.07; after vs. before resident duty hour reform). Conclusion Resident duty hour restrictions were associated with a reduction in the number of yearly hours worked by residents in the NICU as well as a significant decrease in adjusted odds of early mortality but not of hospital mortality in admitted neonates.

scholarly journals Clinical characteristics and outcomes of neonates with polymicrobial ventilator-associated pneumonia in the intensive care unit

2021 ◽  
Author(s):  
Hsiao-Chin Wang ◽  
Ming-Horng Tsai ◽  
Shih-Ming Chu ◽  
Chen-Chu Liao ◽  
Mei-Yin Lai ◽  
...  
Keyword(s):  
Intensive Care Unit ◽  
Intensive Care ◽  
Neonatal Intensive Care Unit ◽  
Treatment Failure ◽  
Clinical Features ◽  
Neonatal Intensive Care ◽  
Clinical Characteristics ◽  
Ventilator Associated Pneumonia ◽  
Prolonged Intubation ◽  
Therapeutic Responses

Abstract Background: Ventilator associated pneumonia (VAP) caused by more than one microorganisms is not uncommon and may be potentially challenging, but the relevant data is scarce in ventilated neonates. We aimed to investigate the clinical characteristics and outcomes of polymicrobial VAP in the neonatal intensive care unit (NICU).Methods: All neonates with definite diagnosis of VAP from a tertiary level neonatal intensive care unit (NICU) in Taiwan between October 2017 and September 2020 were prospectively observed and enrolled for analyses. All clinical features, therapeutic interventions and outcomes were compared between the polymicrobial VAP and monomicrobial VAP episodes. Multivariate regression analyses were used to find the independent risk factors for treatment failure. Results: Among 236 episodes of neonatal VAP, 60 (25.4%) were caused by more than one microorganisms. Polymicrobial VAP episodes were more likely to be associated with multidrug-resistant pathogens (53.3% versus 34.7%, P = 0.014), more often occurred in later days of life and in neonates with prolonged intubation and underlying bronchopulmonary dysplasia. Otherwise most clinical characteristics of polymicrobial VAP were similar to those of monomicrobial VAP. The therapeutic responses and treatment outcomes were also comparable between these two groups, although modification of therapeutic antibiotics were significantly more common in polymicrobial VAP episodes than monomicrobial VAP episodes (63.3% versus 46.2%; P<0.001). None of any specific pathogens was significantly associated with worse outcomes. Instead, it is the severity of illness, including presence of concurrent bacteremia, septic shock, and requirement of high-frequency oscillatory ventilator and underlying neurological sequelae that are independently associated with treatment failure.Conclusions: Polymicrobial VAP accounted for 25.4% of all neonatal VAP in the NICU, and frequently occurred in neonates with prolonged intubation and underlying bronchopulmonary dysplasia. In our cohort, most clinical features, therapeutic responses and final outcomes of neonates with monomicrobial and polymicrobial VAP did not differ significantly.

Sign in / Sign up

Export Citation Format

Share Document