Artificial Intelligence and Machine Learning New Techniques for Generative AI that Drive Outcomes for Health and Transportation

Artificial Intelligence (AI) and Machine Learning (ML) have emerged as powerful tools for data-driven decisionmaking across critical sectors such as healthcare and transportation. However, traditional ML models often struggle with data imbalance and scarcity, leading to biased predictions and limited generalization. This study introduces an advanced Generative AI-driven hybrid framework that integrates a Tabular Generative Adversarial Network (TGAN) with the XGBoost classifier to overcome these limitations and enhance predictive reliability. The framework is validated using two benchmark datasets: the heart-disease dataset for binary diagnosis in the healthcare application and the UK roadaccident dataset for multi-class severity prediction in the transportation application. The Generative AI component learns the statistical patterns of real data and generates highquality synthetic records for under-represented cases such as disease-positive patients and fatal accidents. These synthetic samples are merged with the real training data to form a balanced, information-rich dataset, while the test set remains unaltered to ensure unbiased evaluation. Experimental findings confirm that Generative AI augmentation significantly enhances predictive reliability. The TGAN-XGBoost model achieved an accuracy of$\mathbf{9 8. 9 2 \%}$for heart-disease classification and 98.26% for accident-severity prediction, demonstrating substantial improvement over baseline results. The generated data closely mirror real distributions, ensuring that model gains stem from genuine representational enrichment rather than overfitting. This research highlights the transformative role of Generative AI in augmenting traditional ML workflows, offering a scalable, domain-independent strategy for achieving robust, data-balanced, and high-accuracy outcomes in both health diagnostics and transportation safety applications.