Artificial Intelligence and Machine Learning Models in the Prediction, Analysis, and Treatment Support of Rotavirus and Pediatric Diarrhea: A Systematic Review (2020- 2025)

Rotavirus is a viral infection that affect mostly children below 5 years of age. It remains a leading cause of acute gastroenteritis and diarrheal morbidity among children under five years globally, despite vaccine availability. It causes inflammation of the digestive traits which results in diarrhea, vomiting and fever. With the advent of Artificial Intelligence (AI) and Machine Learning (ML) models, so many models in these areas have been applied to improve early diagnosis, genotype classification, outbreak forecasting, risk stratification, and clinical decision support for rotavirus and pediatric diarrhea diseases. This systematic review from (2020-2025) synthesizes evidence from recent studies and publications across various journals employing supervised machine learning, deep learning (DL), hybrid architectures, ensemble methods, and integrative prediction frameworks in the prediction, diagnosis and treatment support of Rotavirus and Pediatrics diarrhea Across clinical, genomic, epidemiological, and environmental datasets, Random Forest (RF) models consistently demonstrated strong predictive performance, while hybrid deep learning architectures such as VGG–DenseNet combinations achieved high diagnostic accuracy with improved interpretability. AI-based genomic classification models achieved near-perfect genotype identification, supporting surveillance and vaccine strategy. Forecasting models integrating meteorological and seasonal variables outperformed traditional statistical approaches such as ARIMA. Despite promising results, limitations persist; the problem of small datasets leading to overfitting of models, lack of external validation, class imbalance, limited generalizability across regions due to regional peculiarities and factors and underrepresentation of low-resource settings has become nuancing factors. Future research should prioritize multimodal learning, federated frameworks, large-scale validation, and integration into clinical workflows in sub-Saharan Africa and other high-burden regions to achieve optimal results.