This milestone validates the successful design, optimization, and internal testing of three innovative federated learning methods: Centre Dropout, Weight Smoothing, and Uncertainty Awareness, demonstrating their readiness for deployment within the DataTools4Heart consortium. Centre Dropout offers a practical solution to improve training efficiency and fairness across heterogeneous healthcare datasets by selectively excluding centres and proportionally adjusting contributions without sacrificing predictive performance. Weight Smoothing addresses aggregation bias towards data-rich centres, showing consistent gains in federated settings compared to local models, with more significant effects anticipated on diverse external datasets. The Uncertainty-Aware Federated Learning approach effectively incorporates prediction confidence by weighting model updates based on uncertainty, providing a novel mechanism to enhance robustness in clinical AI applications. Future work will focus on extending validation to real-world DT4H datasets, ensuring practical applicability in cross- institutional medical data integration.