A Systematic Review of Machine Learning Applications in the Understanding of Early Childhood Adversity
1
Epidemiology, Charles Darwin University, Menzies School of Health research, Darwin, Australia
Popul. Med. 2026;8(Supplement Supplement 1):
ABSTRACT
ABSTRACT:
Background / Research
OBJECTIVES:
Machine learning (ML) is an emerging approach to investigating childhood adversity. This systematic review critically examines the ML methodologies utilised to understand Early Childhood Adversity (ECA) in children from birth to age six over the past decade. This review aims to evaluate the types of ML employed, the performance of ML algorithms, and gather relevant and reliable information to synthesise existing evidence.
METHODS:
A comprehensive literature search was conducted using PubMed, Scopus, Web of Science, APA PsycINFO, MEDLINE, and CINAHL databases. In addition, Google Scholar and citation search from a list of eligible studies were included. The protocol has been registered in the PROSPERO (protocol No. CRD42024591136). Narrative synthesis was used to present the study characteristics, context, and findings. We assessed the methodological quality of research using the Prediction Model Risk of Bias Assessment Tool (PROBAST) and compiled the results in both descriptive and tabular formats.
RESULTS:
A total of 892 studies were retrieved from electronic databases, of which 35 full-text articles were assessed in detail. Finally, eight studies met the inclusion criteria and were included in the final review. The studies utilised supervised, unsupervised, semi-supervised, and ensemble techniques for model development to address both prediction and clustering analysis. Random Forests, Support Vector Machines, and K-means clustering were the most utilised for prediction and clustering of ECA.
CONCLUSIONS:
In conclusion, the review synthesised findings from eight studies to explore how ML has been applied in ECA studies, indicating that ML models can predict ECA and cluster cases based on shared characteristics. The review reveals that the overall evidence remains limited, highlighting the need for further research. Future research should prioritise robust preprocessing, appropriate feature selection, and algorithms with dimensionality reduction to enhance accuracy, interpretability, and generalisability.
Background / Research
OBJECTIVES:
Machine learning (ML) is an emerging approach to investigating childhood adversity. This systematic review critically examines the ML methodologies utilised to understand Early Childhood Adversity (ECA) in children from birth to age six over the past decade. This review aims to evaluate the types of ML employed, the performance of ML algorithms, and gather relevant and reliable information to synthesise existing evidence.
METHODS:
A comprehensive literature search was conducted using PubMed, Scopus, Web of Science, APA PsycINFO, MEDLINE, and CINAHL databases. In addition, Google Scholar and citation search from a list of eligible studies were included. The protocol has been registered in the PROSPERO (protocol No. CRD42024591136). Narrative synthesis was used to present the study characteristics, context, and findings. We assessed the methodological quality of research using the Prediction Model Risk of Bias Assessment Tool (PROBAST) and compiled the results in both descriptive and tabular formats.
RESULTS:
A total of 892 studies were retrieved from electronic databases, of which 35 full-text articles were assessed in detail. Finally, eight studies met the inclusion criteria and were included in the final review. The studies utilised supervised, unsupervised, semi-supervised, and ensemble techniques for model development to address both prediction and clustering analysis. Random Forests, Support Vector Machines, and K-means clustering were the most utilised for prediction and clustering of ECA.
CONCLUSIONS:
In conclusion, the review synthesised findings from eight studies to explore how ML has been applied in ECA studies, indicating that ML models can predict ECA and cluster cases based on shared characteristics. The review reveals that the overall evidence remains limited, highlighting the need for further research. Future research should prioritise robust preprocessing, appropriate feature selection, and algorithms with dimensionality reduction to enhance accuracy, interpretability, and generalisability.
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
