Dataset Information

Data-driven longitudinal characterization of neonatal health and morbidity.

ABSTRACT: Although prematurity is the single largest cause of death in children under 5 years of age, the current definition of prematurity, based on gestational age, lacks the precision needed for guiding care decisions. Here, we propose a longitudinal risk assessment for adverse neonatal outcomes in newborns based on a deep learning model that uses electronic health records (EHRs) to predict a wide range of outcomes over a period starting shortly before conception and ending months after birth. By linking the EHRs of the Lucile Packard Children's Hospital and the Stanford Healthcare Adult Hospital, we developed a cohort of 22,104 mother-newborn dyads delivered between 2014 and 2018. Maternal and newborn EHRs were extracted and used to train a multi-input multitask deep learning model, featuring a long short-term memory neural network, to predict 24 different neonatal outcomes. An additional cohort of 10,250 mother-newborn dyads delivered at the same Stanford Hospitals from 2019 to September 2020 was used to validate the model. Areas under the receiver operating characteristic curve at delivery exceeded 0.9 for 10 of the 24 neonatal outcomes considered and were between 0.8 and 0.9 for 7 additional outcomes. Moreover, comprehensive association analysis identified multiple known associations between various maternal and neonatal features and specific neonatal outcomes. This study used linked EHRs from more than 30,000 mother-newborn dyads and would serve as a resource for the investigation and prediction of neonatal outcomes. An interactive website is available for independent investigators to leverage this unique dataset: https://maternal-child-health-associations.shinyapps.io/shiny_app/.

SUBMITTER: De Francesco D

PROVIDER: S-EPMC10197092 | biostudies-literature | 2023 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Data-driven longitudinal characterization of neonatal health and morbidity.

De Francesco Davide D Reiss Jonathan D JD Roger Jacquelyn J Tang Alice S AS Chang Alan L AL Becker Martin M Phongpreecha Thanaphong T Espinosa Camilo C Morin Susanna S Berson Eloïse E Thuraiappah Melan M Le Brian L BL Ravindra Neal G NG Payrovnaziri Seyedeh Neelufar SN Mataraso Samson S Kim Yeasul Y Xue Lei L Rosenstein Melissa G MG Oskotsky Tomiko T Marić Ivana I Gaudilliere Brice B Carvalho Brendan B Bateman Brian T BT Angst Martin S MS Prince Lawrence S LS Blumenfeld Yair J YJ Benitz William E WE Fuerch Janene H JH Shaw Gary M GM Sylvester Karl G KG Stevenson David K DK Sirota Marina M Aghaeepour Nima N

Science translational medicine 20230215 683

Although prematurity is the single largest cause of death in children under 5 years of age, the current definition of prematurity, based on gestational age, lacks the precision needed for guiding care decisions. Here, we propose a longitudinal risk assessment for adverse neonatal outcomes in newborns based on a deep learning model that uses electronic health records (EHRs) to predict a wide range of outcomes over a period starting shortly before conception and ending months after birth. By linki ...[more]

PMID: 36791208

Similar Datasets

Project description:BackgroundNeonatal morbidity is associated with lifelong impairments, but the absence of a consensual definition and the need for large data sets limit research.ObjectivesTo inform initiatives to define standard outcomes for research, we reviewed composite neonatal morbidity indicators derived from routine hospital discharge data.Data sourcesPubMed (updated on October 12, 2018). The search algorithm was based on three components: "morbidity," "neonatal," and "hospital discharge data."Study selection and data extractionStudies investigating neonatal morbidity using a composite indicator based on hospital discharge data were included. Indicators defined for specific conditions (eg congenital anomalies, maternal addictions) were excluded. The target population, objectives, component morbidities, diagnosis and procedure codes, validation methods, and prevalence of morbidity were extracted.SynthesisFor each study, we assessed construct validity by describing the methods used to select the indicator components and evaluated whether the authors assessed internal and external validity. We also calculated confidence intervals for the prevalence of the morbidity composite.ResultsSeventeen studies fulfilled inclusion criteria. Indicators targeted all (n = 4), low-/moderate-risk (n = 9), and very preterm (VPT, n = 4) infants. Components were similar for VPT infants, but domains and diagnosis codes within domains varied widely for all and low-/moderate-risk infants. Component selection was described for 8/17 indicators and some form of validation reported for 12/17. Neonatal morbidity prevalence ranged from 4.6% to 9.0% of all infants, 0.4% to 8.0% of low-/moderate-risk infants, and 17.8% to 61.0% of VPT infants.ConclusionsMultiple neonatal morbidity indicators based on hospital discharge data have been used for research, but their heterogeneity limits comparisons between studies. Standard neonatal outcome measures are needed for benchmarking and synthesis of research results.

Project description:Background: The Gambia Demographic and Health Survey 2013 data showed that up to 63% of deliveries in the country occur in health facilities. Despite such a high rate, there are few facility-based studies on delivery outcomes in the country. This analysis ancillary to a randomized control trial describes occurrence of poor pregnancy outcomes in a cohort of women and their infants delivering in a government health facility in urban Gambia. Methods: Using clinical information obtained during the trial, we calculated rates of poor pregnancy outcomes including stillbirths, hospitalization and neonatal deaths. Logistic regression was used to calculate odds ratio (OR) and 95% confidence interval (CI) in the risk factors analysis. Results: Between April 2013 and 2014, 829 mothers delivered 843 babies, including 13 stillbirths [15.4 (7.1-23.8)] per 1,000 births. Among 830 live born infants, 7.6% (n = 63) required hospitalization during the 8-week follow-up period. Most of these hospitalizations (74.6%) occurred during the early neonatal period (<7 days of life). Severe clinical infections (i.e., sepsis, meningitis and pneumonia) (n = 27) were the most common diagnoses, followed by birth asphyxia (n = 13), major congenital malformations (n = 10), jaundice (n = 6) and low birth weight (n = 5). There were sixteen neonatal deaths, most of which also occurred during the early neonatal period. Overall, neonatal mortality rate (NMR) and perinatal mortality rate (PMR) were 19.3 (CI: 9.9-28.7) per 1,000 live births and 26.1 (CI: 15.3-36.9) per 1,000 total births, respectively. Severe clinical infections and birth asphyxia accounted for 37 and 31% of neonatal deaths, respectively. The risk of hospitalization was higher among neonates with severe congenital malformations, low birth weight, twin deliveries, and those born by cesarean section. Risk of mortality was higher among neonates with severe congenital malformations and twin deliveries. Conclusion: Neonatal hospitalization and deaths in our cohort were high. Although vertical interventions may reduce specific causes of morbidity and mortality, data indicate the need for a holistic approach to significantly improve the rates of poor pregnancy outcomes. Critically, a focus on decreasing the high rate of stillbirths is warranted. Clinical Trial Registration: ClinicalTrials.gov Identifier: NCT01800942.

Project description:BackgroundSocial class is frequently used as a means of ranking the population to expose inequalities in health, but less often as a means of understanding the social processes of causation. We explored how effectively different social class mechanisms could be measured by longitudinal cohort data and whether those measures were able to explain health outcomes.MethodsUsing a theoretically informed approach, we sought to map variables within the National Child Development Study (NCDS) to five different social class mechanisms: social background and early life circumstances; habitus and distinction; exploitation and domination; location within market relations; and power relations. Associations between the SF-36 physical, emotional and general health outcomes at age 50 years and the social class measures within NCDS were then assessed through separate multiple linear regression models. R2 values were used to quantify the proportion of variance in outcomes explained by the independent variables.ResultsWe were able to map the NCDS variables to the each of the social class mechanisms except 'Power relations'. However, the success of the mapping varied across mechanisms. Furthermore, although relevant associations between exposures and outcomes were observed, the mapped NCDS variables explained little of the variation in health outcomes: for example, for physical functioning, the R2 values ranged from 0.04 to 0.10 across the four mechanisms we could map.ConclusionsThis study has demonstrated both the potential and the limitations of available cohort studies in measuring aspects of social class theory. The relatively small amount of variation explained in the outcome variables in this study suggests that these are imperfect measures of the different social class mechanisms. However, the study lays an important foundation for further research to understand the complex interactions, at various life stages, between different aspects of social class and subsequent health outcomes.

Project description:BackgroundLong-term survival after premature birth is significantly determined by development of morbidities, primarily affecting the cardio-respiratory or central nervous system. Existing studies are limited to pairwise morbidity associations, thereby lacking a holistic understanding of morbidity co-occurrence and respective risk profiles.MethodsOur study, for the first time, aimed at delineating and characterizing morbidity profiles at near-term age and investigated the most prevalent morbidities in preterm infants: bronchopulmonary dysplasia (BPD), pulmonary hypertension (PH), mild cardiac defects, perinatal brain pathology and retinopathy of prematurity (ROP). For analysis, we employed two independent, prospective cohorts, comprising a total of 530 very preterm infants: AIRR ("Attention to Infants at Respiratory Risks") and NEuroSIS ("Neonatal European Study of Inhaled Steroids"). Using a data-driven strategy, we successfully characterized morbidity profiles of preterm infants in a stepwise approach and (1) quantified pairwise morbidity correlations, (2) assessed the discriminatory power of BPD (complemented by imaging-based structural and functional lung phenotyping) in relation to these morbidities, (3) investigated collective co-occurrence patterns, and (4) identified infant subgroups who share similar morbidity profiles using machine learning techniques.ResultsFirst, we showed that, in line with pathophysiologic understanding, BPD and ROP have the highest pairwise correlation, followed by BPD and PH as well as BPD and mild cardiac defects. Second, we revealed that BPD exhibits only limited capacity in discriminating morbidity occurrence, despite its prevalence and clinical indication as a driver of comorbidities. Further, we demonstrated that structural and functional lung phenotyping did not exhibit higher association with morbidity severity than BPD. Lastly, we identified patient clusters that share similar morbidity patterns using machine learning in AIRR (n=6 clusters) and NEuroSIS (n=8 clusters).ConclusionsBy capturing correlations as well as more complex morbidity relations, we provided a comprehensive characterization of morbidity profiles at discharge, linked to shared disease pathophysiology. Future studies could benefit from identifying risk profiles to thereby develop personalized monitoring strategies.Trial registrationAIRR: DRKS.de, DRKS00004600, 28/01/2013. NEuroSIS: ClinicalTrials.gov, NCT01035190, 18/12/2009.

Project description:PurposeTo evaluate long-term outcome of three years and treatment patterns of patients suffering from severely drug-refractory epilepsy (SDRE).MethodsThis analysis was population-based and retrospective, with data collected from four million individuals insured by statutory German health insurance. ICD-10 codes for epilepsy (G40*) and intake of anticonvulsants were used to identify prevalent cases, which were then compared with a matched cohort drawn from the population at large. Insurance data were available from 2008 to 2013. Any patient who had been prescribed with at least four different antiepileptic drugs (AEDs) in an 18-month period was defined as an SDRE case.ResultsA total of 769 patients with SDRE were identified. Of these, 19% were children and adolescents; the overall mean age was 42.3 years, 45.4% were female and 54.6% male. An average of 2.7 AEDs per patient was prescribed during the first follow-up year. The AEDs most commonly prescribed were: levetiracetam (53.5%), lamotrigine (41.4%), valproate (41.3%), lacosamide (20.4%), and topiramate (17.8%). During 3-year follow-up, there was an annual rate of hospitalization in the range 42.7 to 55%, which was significantly higher than the 11.6-12.8% (p < 0.001) for the matched controls. Admissions to hospital because of epilepsy ranged between 1.7 and 1.9 per year, with an average duration for each epilepsy-caused hospitalization of 10-11.1 days. The number of comorbidities for SDRE patients was significantly increased compared with the matched controls: depression (28% against 10%), vascular disorders (22% against 5%), and injury rates were also higher (head 16% against 3%, trunk and limbs 16% against 8%). The 3-year mortality rate for SDRE patients was 14% against 2.1% in the matched cohort.ConclusionSDRE patients are treated with AED polytherapy for all of the 3-year follow-up period. They are hospitalized more frequently than the general population and show increased morbidity levels and a sevenfold increase in mortality rate over 3 years. Further examination is required of ways in which new approaches to treatment could lead to better outcomes in severely affected patients.

Dataset Information

Data-driven longitudinal characterization of neonatal health and morbidity.

Publications

Data-driven longitudinal characterization of neonatal health and morbidity.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets