Machine learning-based prediction of acute severity in infants hospitalized for bronchiolitis: a multicenter prospective study.
Ontology highlight
ABSTRACT:
SUBMITTER: Raita Y
PROVIDER: S-EPMC7335203 | biostudies-literature | 2020 Jul
REPOSITORIES: biostudies-literature
Publications
Machine learning-based prediction of acute severity in infants hospitalized for bronchiolitis: a multicenter prospective study.
Raita Yoshihiko Y Camargo Carlos A CA Macias Charles G CG Mansbach Jonathan M JM Piedra Pedro A PA Porter Stephen C SC Teach Stephen J SJ Hasegawa Kohei K
Scientific reports 20200703 1
We aimed to develop machine learning models to accurately predict bronchiolitis severity, and to compare their predictive performance with a conventional scoring (reference) model. In a 17-center prospective study of infants (aged < 1 year) hospitalized for bronchiolitis, by using routinely-available pre-hospitalization data as predictors, we developed four machine learning models: Lasso regression, elastic net regression, random forest, and gradient boosted decision tree. We compared their pred ...[more]
Similar Datasets
Project description:ObjectiveTo investigate the relationship between socioeconomic factors and bronchiolitis severity among hospitalized infants.MethodsWe performed a 17-center, prospective cohort study from 2011 to 2014. Children <1 year old hospitalized with bronchiolitis were enrolled. Socioeconomic factors included estimated median household income (MHI) per home ZIP code, parent-reported household income, number of adults and children in household, and insurance type. We defined higher bronchiolitis severity as receipt of intensive care treatment. Multivariable logistic regression was used to analyze the association between socioeconomic factors and bronchiolitis severity, with the final model adjusted for potential clustering by site.ResultsIn multivariable models adjusted for demographic and clinical characteristics, estimated MHI was the socioeconomic factor most strongly associated with severity. Compared to infants with an intermediate MHI ($40,000-$79,999), odds of receiving intensive care treatment were significantly higher for those with MHI of ≥$80,000 (aOR 2.05, 95% CI 1.19-3.53). No significant associations were found for the other socioeconomic factors (all P > .30). While there were no significant differences in clinical presentation between income groups (all P > .25) or in receipt of mechanical ventilation alone (P = .98), infants with estimated MHI ≥$80,000 were significantly more likely to specifically have been admitted to the intensive care unit (P = .01).ConclusionsIn this multicenter study of infants hospitalized with bronchiolitis, we identified higher median household income as a risk factor for intensive care treatment. This work may yield important biological or nonbiological insights for the future management of infants with bronchiolitis.
| S-EPMC6930979 | biostudies-literature
Project description:Acute bronchiolitis treatment in children and infants is largely supportive, but chest physiotherapy is routinely performed in some countries. In France, national guidelines recommend a specific type of physiotherapy combining the increased exhalation technique (IET) and assisted cough (AC). Our objective was to evaluate the efficacy of chest physiotherapy (IET + AC) in previously healthy infants hospitalized for a first episode of acute bronchiolitis.We conducted a multicenter, randomized, outcome assessor-blind and parent-blind trial in seven French pediatric departments. We recruited 496 infants hospitalized for first-episode acute bronchiolitis between October 2004 and January 2008. Patients were randomly allocated to receive from physiotherapists three times a day, either IET + AC (intervention group, n=246) or nasal suction (NS, control group, n=250). Only physiotherapists were aware of the allocation group of the infant. The primary outcome was time to recovery, defined as 8 hours without oxygen supplementation associated with minimal or no chest recession, and ingesting more than two-thirds of daily food requirements. Secondary outcomes were intensive care unit admissions, artificial ventilation, antibiotic treatment, description of side effects during procedures, and parental perception of comfort. Statistical analysis was performed on an intent-to-treat basis. Median time to recovery was 2.31 days, (95% confidence interval [CI] 1.97-2.73) for the control group and 2.02 days (95% CI 1.96-2.34) for the intervention group, indicating no significant effect of physiotherapy (hazard ratio [HR]=1.09, 95% CI 0.91-1.31, p=0.33). No treatment by age interaction was found (p=0.97). Frequency of vomiting and transient respiratory destabilization was higher in the IET + AC group during the procedure (relative risk [RR]=10.2, 95% CI 1.3-78.8, p=0.005 and RR=5.4, 95% CI 1.6-18.4, p=0.002, respectively). No difference between groups in bradycardia with or without desaturation (RR=1.0, 95% CI 0.2-5.0, p=1.00 and RR=3.6, 95% CI 0.7-16.9, p=0.10, respectively) was found during the procedure. Parents reported that the procedure was more arduous in the group treated with IET (mean difference=0.88, 95% CI 0.33-1.44, p=0.002), whereas there was no difference regarding the assessment of the child's comfort between both groups (mean difference=-0.07, 95% CI -0.53 to 0.38, p=0.40). No evidence of differences between groups in intensive care admission (RR=0.7, 95% CI 0.3-1.8, p=0.62), ventilatory support (RR=2.5, 95% CI 0.5-13.0, p=0.29), and antibiotic treatment (RR=1.0, 95% CI 0.7-1.3, p=1.00) was observed.IET + AC had no significant effect on time to recovery in this group of hospitalized infants with bronchiolitis. Additional studies are required to explore the effect of chest physiotherapy on ambulatory populations and for infants without a history of atopy.ClinicalTrials.gov NCT00125450.
| S-EPMC2946956 | biostudies-literature
Project description:BackgroundBronchiolitis is the leading cause of infant hospitalizations in the United States. Growing evidence supports the heterogeneity of bronchiolitis. However, little is known about the interrelationships between major respiratory viruses (and their species), host systemic metabolism, and disease pathobiology.MethodsIn an ongoing multicenter prospective cohort study, we profiled the serum metabolome in 113 infants (63 RSV-only, 21 RV-A, and 29 RV-C) hospitalized with bronchiolitis. We identified serum metabolites that are most discriminatory in the RSV-RV-A and RSV-RV-C comparisons using sparse partial least squares discriminant analysis. We then investigated the association between discriminatory metabolites with acute and chronic outcomes.ResultsIn 113 infants with bronchiolitis, we measured 639 metabolites. Serum metabolomic profiles differed in both comparisons (Ppermutation < 0.05). In the RSV-RV-A comparison, we identified 30 discriminatory metabolites, predominantly in lipid metabolism pathways (eg, sphingolipids and carnitines). In multivariable models, these metabolites were significantly associated with the risk of clinical outcomes (eg, tricosanoyl sphingomyelin, OR for recurrent wheezing at age of 3 years = 1.50; 95% CI: 1.05-2.15). In the RSV-RV-C comparison, the discriminatory metabolites were also primarily involved in lipid metabolism (eg, glycerophosphocholines [GPCs], 12,13-diHome). These metabolites were also significantly associated with the risk of outcomes (eg, 1-stearoyl-2-linoleoyl-GPC, OR for positive pressure ventilation use during hospitalization = 0.47; 95% CI: 0.28-0.78).ConclusionRespiratory viruses and their species had distinct serum metabolomic signatures that are associated with differential risks of acute and chronic morbidities of bronchiolitis. Our findings advance research into the complex interrelations between viruses, host systemic response, and bronchiolitis pathobiology.
| S-EPMC7704725 | biostudies-literature
Project description:ObjectiveMachine learning models trained on electronic health records have achieved high prognostic accuracy in test datasets, but little is known about their embedding into clinical workflows. We implemented a random forest-based algorithm to identify hospitalized patients at high risk for delirium, and evaluated its performance in a clinical setting.Materials and methodsDelirium was predicted at admission and recalculated on the evening of admission. The defined prediction outcome was a delirium coded for the recent hospital stay. During 7 months of prospective evaluation, 5530 predictions were analyzed. In addition, 119 predictions for internal medicine patients were compared with ratings of clinical experts in a blinded and nonblinded setting.ResultsDuring clinical application, the algorithm achieved a sensitivity of 74.1% and a specificity of 82.2%. Discrimination on prospective data (area under the receiver-operating characteristic curve = 0.86) was as good as in the test dataset, but calibration was poor. The predictions correlated strongly with delirium risk perceived by experts in the blinded (r = 0.81) and nonblinded (r = 0.62) settings. A major advantage of our setting was the timely prediction without additional data entry.DiscussionThe implemented machine learning algorithm achieved a stable performance predicting delirium in high agreement with expert ratings, but improvement of calibration is needed. Future research should evaluate the acceptance of implemented machine learning algorithms by health professionals.ConclusionsOur study provides new insights into the implementation process of a machine learning algorithm into a clinical workflow and demonstrates its predictive power for delirium.
| S-EPMC7647341 | biostudies-literature
Project description:We reexamined the finding of an inverse relationship between values of nasopharyngeal lactate dehydrogenase, a marker of the innate immune response, and bronchiolitis severity. In a prospective, multicenter study of 258 children, we found in a multivariable model that higher nasopharyngeal lactate dehydrogenase values in young children with bronchiolitis were independently associated with a decreased risk of hospitalization.
| S-EPMC3375381 | biostudies-literature
Project description:Gene expression profiles were generated from 199 primary breast cancer patients. Samples 1-176 were used in another study, GEO Series GSE22820, and form the training data set in this study. Sample numbers 200-222 form a validation set. This data is used to model a machine learning classifier for Estrogen Receptor Status. RNA was isolated from 199 primary breast cancer patients. A machine learning classifier was built to predict ER status using only three gene features.
2013-01-01 | E-GEOD-29210 | biostudies-arrayexpress
Project description:Background: There is limited data on how different RSV genotypes and associated viral loads influence disease phenotypes. We characterized the genetic variability of RSV strains during five non-consecutive respiratory seasons, and evaluated the role of RSV subtypes, genotypes and viral loads on clinical disease severity. Methods: Healthy infants hospitalized with RSV bronchiolitis were prospectively enrolled and nasopharyngeal samples obtained within 24h of hospitalization for RSV load quantitation by PCR, typing and genotyping. Parameters of disease severity were assessed, and multivariate models constructed to identify virologic and clinical factors predictive of clinical outcomes. Results: From March 2004 to April 2011, we enrolled 253 patients (56.5 % males; median age 2.1 (1.1-4.0) months). RSV A infections predominated over RSV B (69% vs. 31%; p<0.001) and showed greater genotype variability. The most common genotypes were RSV A/GA2, A/GA5 and RSV B/BA. Infants infected with RSV GA5 had higher viral loads compared with GA2 or BA infection (p<0.01), independent of duration of symptoms. After adjusting for other covariates, RSV A/GA5 infections were associated with longer hospital stay. Conclusions: RSV A infections were more frequent than RSV B infections and displayed greater genetic variability. Infections with GA5 were independently associated with clinical disease severity.
2018-03-21 | GSE103842 | GEO
Project description:Little is known about the relationship between the specific airway microbiota composition and severity of bronchiolitis. We aimed to identify nasopharyngeal microbiota profiles and link these profiles to acute severity in infants hospitalised for bronchiolitis.We conducted a multicentre prospective cohort study of 1005 infants (age <1?year) hospitalised for bronchiolitis over three winters, 2011-2014. By applying a 16S rRNA gene sequence and clustering approach to the nasopharyngeal aspirates collected within 24?h of hospitalisation, we determined nasopharyngeal microbiota profiles and their association with bronchiolitis severity. The primary outcome was intensive care use, i.e. admission to an intensive care unit or use of mechanical ventilation.We identified four nasopharyngeal microbiota profiles: three profiles were dominated by one of Haemophilus, Moraxella or Streptococcus, while the fourth profile had the highest bacterial richness. The rate of intensive care use was highest in infants with a Haemophilus-dominant profile and lowest in those with a Moraxella-dominant profile (20.2% versus 12.3%; unadjusted OR 1.81, 95% CI 1.07-3.11, p=0.03). After adjusting for 11 patient-level confounders, the rate remained significantly higher in infants with Haemophilus-dominant profiles (OR 1.98, 95% CI 1.08-3.62, p=0.03). These findings were externally validated in a separate cohort of 307 children hospitalised for bronchiolitis.
| S-EPMC5459592 | biostudies-literature
Project description:Recent resting-state functional MRI investigations have demonstrated that much of the large-scale functional network architecture supporting motor, sensory and cognitive functions in older pediatric and adult populations is present in term- and prematurely-born infants. Application of new analytical approaches can help translate the improved understanding of early functional connectivity provided through these studies into predictive models of neurodevelopmental outcome. One approach to achieving this goal is multivariate pattern analysis, a machine-learning, pattern classification approach well-suited for high-dimensional neuroimaging data. It has previously been adapted to predict brain maturity in children and adolescents using structural and resting state-functional MRI data. In this study, we evaluated resting state-functional MRI data from 50 preterm-born infants (born at 23-29weeks of gestation and without moderate-severe brain injury) scanned at term equivalent postmenstrual age compared with data from 50 term-born control infants studied within the first week of life. Using 214 regions of interest, binary support vector machines distinguished term from preterm infants with 84% accuracy (p<0.0001). Inter- and intra-hemispheric connections throughout the brain were important for group categorization, indicating that widespread changes in the brain's functional network architecture associated with preterm birth are detectable by term equivalent age. Support vector regression enabled quantitative estimation of birth gestational age in single subjects using only term equivalent resting state-functional MRI data, indicating that the present approach is sensitive to the degree of disruption of brain development associated with preterm birth (using gestational age as a surrogate for the extent of disruption). This suggests that support vector regression may provide a means for predicting neurodevelopmental outcome in individual infants.
| S-EPMC4914443 | biostudies-literature
Project description:ObjectiveTo compare machine learning methods for predicting inpatient seizures risk and determine the feasibility of 1-h screening EEG to identify low-risk patients (<5% seizures risk in 48 h).MethodsThe Critical Care EEG Monitoring Research Consortium (CCEMRC) multicenter database contains 7716 continuous EEGs (cEEG). Neural networks (NN), elastic net logistic regression (EN), and sparse linear integer model (RiskSLIM) were trained to predict seizures. RiskSLIM was used previously to generate 2HELPS2B model of seizure predictions. Data were divided into training (60% for model fitting) and evaluation (40% for model evaluation) cohorts. Performance was measured using area under the receiver operating curve (AUC), mean risk calibration (CAL), and negative predictive value (NPV). A secondary analysis was performed using Monte Carlo simulation (MCS) to normalize all EEG recordings to 48 h and use only the first hour of EEG as a "screening EEG" to generate predictions.ResultsRiskSLIM recreated the 2HELPS2B model. All models had comparable AUC: evaluation cohort (NN: 0.85, EN: 0.84, 2HELPS2B: 0.83) and MCS (NN: 0.82, EN; 0.82, 2HELPS2B: 0.81) and NPV (absence of seizures in the group that the models predicted to be low risk): evaluation cohort (NN: 97%, EN: 97%, 2HELPS2B: 97%) and MCS (NN: 97%, EN: 99%, 2HELPS2B: 97%). 2HELPS2B model was able to identify the largest proportion of low-risk patients.InterpretationFor seizure risk stratification of hospitalized patients, the RiskSLIM generated 2HELPS2B model compares favorably to the complex NN and EN generated models. 2HELPS2B is able to accurately and quickly identify low-risk patients with only a 1-h screening EEG.
| S-EPMC6649418 | biostudies-literature