Background: The inability to objectively diagnose childhood asthma before age five often results in both under‐treatment and over‐treatment of asthma in preschool children. Prediction tools for estimating a child's risk of developing asthma by school‐age could assist physicians in early asthma care for preschool children. This review aimed to systematically identify and critically appraise studies which either developed novel or updated existing prediction models for predicting school‐age asthma.Methods: Three databases (MEDLINE, Embase and Web of Science Core Collection) were searched up to July 2019 to identify studies utilizing information from children ≤5 years of age to predict asthma in school‐age children (6‐13 years). Validation studies were evaluated as a secondary objective.Results: Twenty‐four studies describing the development of 26 predictive models published between 2000 and 2019 were identified. Models were either regression‐based (n = 21) or utilized machine learning approaches (n = 5). Nine studies conducted validations of six regression‐based models. Fifteen (out of 21) models required additional clinical tests. Overall model performance, assessed by area under the receiver operating curve (AUC), ranged between 0.66 and 0.87. Models demonstrated moderate ability to either rule in or rule out asthma development, but not both. Where external validation was performed, models demonstrated modest generalizability (AUC range: 0.62‐0.83).Conclusion: Existing prediction models demonstrated moderate predictive performance, often with modest generalizability when independently validated. Limitations of traditional methods have shown to impair predictive accuracy and resolution. Exploration of novel methods such as machine learning approaches may address these limitations for future school‐age asthma prediction.

Background: Respiratory symptoms are common in early life and often transient. It is difficult to identify in which children these will persist and result in asthma. Machine learning (ML) approaches have the potential for better predictive performance and generalisability over existing childhood asthma prediction models. This study applied ML approaches to predict school‐age asthma (age 10) in early life (Childhood Asthma Prediction in Early life, CAPE model) and at preschool age (Childhood Asthma Prediction at Preschool age, CAPP model).Methods: Clinical and environmental exposure data was collected from children enrolled in the Isle of Wight Birth Cohort (N = 1368, ∼15% asthma prevalence). Recursive Feature Elimination (RFE) identified an optimal subset of features predictive of school‐age asthma for each model. Seven state‐of‐the‐art ML classification algorithms were used to develop prognostic models. Training was performed by applying fivefold cross‐validation, imputation, and resampling. Predictive performance was evaluated on the test set. Models were further externally validated in the Manchester Asthma and Allergy Study (MAAS) cohort.Results: RFE identified eight and twelve predictors for the CAPE and CAPP models, respectively. Support Vector Machine (SVM) algorithms provided the best performance for both the CAPE (area under the receiver operating characteristic curve, AUC = 0.71) and CAPP (AUC = 0.82) models. Both models demonstrated good generalisability in MAAS (CAPE 8‐year = 0.71, 11‐year = 0.71, CAPP 8‐year = 0.83, 11‐year = 0.79) and excellent sensitivity to predict a subgroup of persistent wheezers.Conclusion: Using ML approaches improved upon the predictive performance of existing regression‐based models, with good generalisability and ability to rule in asthma and predict persistent wheeze.

Genome-wide and epigenome-wide association studies have identified genetic variants and differentially methylated nucleotides associated with childhood asthma. Incorporation of such genomic data may improve performance of childhood asthma prediction models which use phenotypic and environmental data. Using genome-wide genotype and methylation data at birth from the Isle of Wight Birth Cohort (n = 1456), a polygenic risk score (PRS), and newborn (nMRS) and childhood (cMRS) methylation risk scores, were developed to predict childhood asthma diagnosis. Each risk score was integrated with two previously published childhood asthma prediction models (CAPE and CAPP) and were validated in the Manchester Asthma and Allergy Study. Individually, the genomic risk scores demonstrated modest-to-moderate discriminative performance (area under the receiver operating characteristic curve, AUC: PRS = 0.64, nMRS = 0.55, cMRS = 0.54), and their integration only marginally improved the performance of the CAPE (AUC: 0.75 vs. 0.71) and CAPP models (AUC: 0.84 vs. 0.82). The limited predictive performance of each genomic risk score individually and their inability to substantially improve upon the performance of the CAPE and CAPP models suggests that genetic and epigenetic predictors of the broad phenotype of asthma are unlikely to have clinical utility. Hence, further studies predicting specific asthma endotypes are warranted.

Micro RNAs (miRNAs) are short, non-coding RNAs (Ribonucleic acids) with regulatory functions that could prove useful as biomarkers for asthma diagnosis and asthma severity-risk stratification. The objective of this systematic review is to identify panels of miRNAs that can be used to support asthma diagnosis and severity-risk assessment. Three databases (Medline, Embase, and SCOPUS) were searched up to 15 September 2020 to identify studies reporting differential expression of specific miRNAs in the tissues of adults and children with asthma. Studies reporting miRNAs associations in animal models that were also studied in humans were included in this review. We identified 75 studies that met our search criteria. Of these, 66 studies reported more than 200 miRNAs that are differentially expressed in asthma patients when compared to non-asthmatic controls. In addition, 16 studies reported 17 miRNAs that are differentially expressed with differences in asthma severity. We were able to construct two panels of miRNAs that are expressed in blood and can serve as core panels to further investigate the practicality and efficiency of using miRNAs as non-invasive biomarkers for asthma diagnosis and severity-risk assessment, respectively.

Several small studies have shown associations between breastfeeding and genome-wide DNA methylation (DNAm). We performed a comprehensive Epigenome-Wide Association Study (EWAS) to identify associations between breastfeeding and DNAm patterns in childhood. We analysed DNAm data from the Isle of Wight Birth Cohort at birth, 10, 18 and 26 years. The feeding method was categorized as breastfeeding duration >3 months and >6 months, and exclusive breastfeeding duration >3 months. EWASs using robust linear regression were performed to identify differentially methylated positions (DMPs) in breastfed and non-breastfed children at age 10 (false discovery rate of 5%). Differentially methylated regions (DMRs) were identified using comb-p. The persistence of significant associations was evaluated in neonates and individuals at 18 and 26 years. Two DMPs, in genes SNX25 and LINC00840, were significantly associated with breastfeeding duration >6 months at 10 years and was replicated for >3 months of exclusive breastfeeding. Additionally, a significant DMR spanning the gene FDFT1 was identified in 10-year-old children who were exposed to a breastfeeding duration >3 months. None of these signals persisted to 18 or 26 years. This study lends further support for a suggestive role of DNAm in the known benefits of breastfeeding on a child’s future health.

Background Wheeze is common in early life and often transient. It is difficult to identify which children will experience persistent symptoms and subsequently develop asthma. Machine learning approaches have the potential for better predictive performance and generalisability over existing childhood asthma prediction models. Objective To apply machine learning approaches for predicting school-age asthma (age 10) in early life (Childhood Asthma Prediction in Early life, CAPE model) and at preschool age (Childhood Asthma Prediction at Preschool age, CAPP model). Methods Data on clinical symptoms and environmental exposures were collected from children enrolled in the Isle of Wight Birth Cohort (N=1368, ∼15% asthma prevalence). Recursive Feature Elimination (RFE) identified the optimal subset of features predictive of school-age asthma for each model. Seven state-of-the-art machine learning classification algorithms were used to develop the models and the results were compared. To optimize the models, training was performed by applying 5-fold cross-validation, imputation and resampling. Predictive performances were evaluated on the test set and externally validated in the Manchester Asthma and Allergy Study (MAAS) cohort. Results RFE identified eight and 12 predictors for the CAPE and CAPP models, respectively. The best predictive performance was demonstrated by a Support Vector Machine (SVM) algorithm for both the CAPE model (area under the receiver operating curve, AUC=0.71) and CAPP model (AUC=0.82). Both models demonstrated good generalisability in MAAS (CAPE 8YR=0.71, 11YR=0.71, CAPP 8YR=0.83, 11YR=0.79). Conclusion Using machine learning approaches improved upon the predictive performance of existing regression-based models, with good generalisability and ability to rule in asthma.