JMIR Publications

ABSTRACT

Background:

The integration of machine learning (ML) in predicting asthma-related outcomes in children presents a novel approach in pediatric healthcare.

Objective:

This scoping review aims to analyze studies since 2019, focusing on ML algorithms, their applications, target populations, and predictive performances.

Methods:

We searched Ovid MEDLINE® ALL and Embase on Ovid, the Cochrane Library (Wiley), CINAHL (Ebsco) and Web of Science (core collection). The search covered the period 2019 to July 18, 2023. Studies applying ML models in predicting asthma-related outcomes in children under 18 years were included. Covidence was used for citation management, and risk of bias (RoB) was assessed using PROBAST.

Results:

From 1,231 initial articles, 15 met our inclusion criteria. The sample size ranged from 74 to 87, 413 patients. The ML techniques studied included logistic regression (n=7, 46.67%), random forests (n=6, 40.00%), gradient boosting (n=4, 26.67%), artificial neural networks (n=3, 20.00%), decision trees (n=2, 13.33%), natural language processing (n=2, 13.33%), and Gaussian mixture model (n=1, 6.67%). The predictive performance, as measured by the Area Under the Curve (AUC), varied across models, with reported ranges from 0.62 to 0.88, indicating differing levels of efficacy. The RoB assessment revealed that a majority of studies (n=8, 53.33%) exhibited low to moderate risk, ensuring a reasonable level of confidence in the findings. Common limitations across studies included data quality issues, sample size constraints, and interpretability concerns.

Conclusions:

This review highlights the diverse application of ML in predicting pediatric asthma outcomes, with each model offering unique strengths and challenges. Future research should address data quality, increase sample sizes, and enhance model interpretability to optimize ML utility in clinical settings for pediatric asthma management.