In a groundbreaking study, researchers delved into the 2016 Ethiopian Demographic and Health Survey dataset to unravel the mysteries surrounding childhood vaccination among children aged 12-23 months. Despite the proven benefits of vaccinations in reducing child mortality, the vaccination coverage in Ethiopia remains low. This study pioneers the use of machine learning algorithms to predict childhood vaccination, shedding light on potential interventions to improve immunisation rates.

The study, embracing a cross-sectional design with a two-stage sampling technique, analysed 1617 samples from the survey dataset. To train and evaluate the model, 70% and 30% of the observations were allocated, respectively. Eight machine learning algorithms were put to the test, with the PART algorithm emerging as the top performer, boasting an impressive 95.53% accuracy.

The researchers identified key attributes influencing childhood vaccination. Antenatal care (ANC) visits, institutional delivery, health facility visits, higher maternal education, and affluent households were highlighted as the top five predictors. These insights provide a roadmap for targeted interventions, emphasising the importance of strengthening ANC services, promoting institutional deliveries, and enhancing maternal education.

The study’s innovation lies in its application of machine learning algorithms to uncover hidden patterns in vaccination data. The researchers used the synthetic minority oversampling technique to address imbalanced data, ensuring a more accurate analysis. Informational gain value guided the selection of crucial attributes, and logical association rules were generated to provide a nuanced understanding of the interplay between various factors influencing childhood vaccination.

Reflecting on the results, the study emphasises the significance of ANC visits, institutional delivery, and maternal education. The findings underscore the need to empower mothers with information during pregnancy, ensuring they are equipped to make informed decisions about childhood vaccination.

The study’s strength lies in its use of nationally representative data, offering insights that could benefit the broader Ethiopian population. While machine learning algorithms lack traditional coefficients, the study’s findings provide a valuable foundation for public health action and decision-making.

In conclusion, the application of machine learning algorithms has unveiled a new frontier in predicting childhood vaccination in Ethiopia. The PART algorithm’s stellar performance, coupled with the identification of crucial attributes, paves the way for targeted interventions. As policymakers and stakeholders digest this research, the hope is that these insights will catalyse efforts to strengthen healthcare systems, improve maternal education, and ultimately enhance childhood vaccination rates across Ethiopia. 

Reference:

Workie, Addisalem & Ayenew, Alex & Walle, Agmasie & Kassie, Sisay & Bekele, Firomsa & Bekana, Teshome. (2023). Machine learning algorithms’ application to predict childhood vaccination among children aged 12-23 months in Ethiopia: Evidence 2016 Ethiopian Demographic and Health Survey dataset. PLOS ONE. 18. 10.1371/journal.pone.0288867. Available from:  https://www.researchgate.net/publication/374811712_Machine_learning_algorithms’_application_to_predict_childhood_vaccination_among_children_aged_12-23_months_in_Ethiopia_Evidence_2016_Ethiopian_Demographic_and_Health_Survey_dataset/citation/download