Hybrid Machine Learning Models For Improving Pediatric Readmission Prediction Using Cloud-Based EMR Analytics
Keywords:
Pediatric Readmissions, Hybrid Machine Learning Models, Cloud-Based Analytics Electronic Medical Records (EMR), Predictive Modeling, Decision Trees, Support Vector Machines (SVM), Neural Networks, Healthcare Data Analysis, Real-Time Predictions, Big Data Analytics, Healthcare Decision Support, Model Accuracy, Readmission Prediction, Data PreprocessingAbstract
Pediatric readmissions place a significant strain on healthcare systems, increasing costs, depleting resources,
and leading to worse health outcomes for children. Traditional predictive models struggle to manage the
complexity and high dimensionality of healthcare data. This study introduces a solution by using hybrid machine
learning models, integrated with cloud-based Electronic Medical Record (EMR) analytics, to more accurately
predict pediatric readmissions. The hybrid model combines decision trees, support vector machines, and neural
networks, effectively capturing complex patterns within healthcare data. Cloud computing’s scalability and
computational power enable real-time processing of large datasets, improving prediction efficiency. The
proposed model outperforms individual algorithms, achieving 87% accuracy, 84% precision, 82% recall, 83%
F1-score, and 91% AUC. These results show that hybrid models on cloud infrastructure provide better predictive
performance, helping healthcare providers make more informed decisions. Additionally, the cloud infrastructure
supports continuous model updates. Future improvements could involve incorporating genomic and
environmental data to enhance prediction accuracy and adaptability.
