Background: Electrocardiogram data, one of the most widely available biosignal data, has become increasingly valuable with the emergence of deep learning methods, providing novel insights into cardiovascular diseases and broader health conditions. However, heterogeneity of electrocardiogram formats, limited access to deep learning model weights, and intricate algorithmic steps for effective fine-tuning for own disease target labels result in complex workflows. Objective: In this work, we introduce ExChanGeAI, a web-based end-to-end platform that unifies and streamlines the ECG analysis workflow – from handling diverse file-types, and interactive visualization, to local, privacy-preserving training and fine-tuning. Methods: The platform offers state-of-the-art deep learning models for training from scratch, and fine-tuning of the provided pre-trained models. ExChanGeAI is adaptable for use on both personal computers and scalable to high performance server environments. Evaluation is conducted for several classification tasks across three external validation sets, including an entirely new testset extracted from routine care. Results: De-novo training with task-specific data outperformed the benchmark foundation model while requiring significantly fewer parameters and lower computational resources. The platform enables users to empirically determine the most suitable model for their specific tasks, based on systematic validations. Conclusion: Conclusions: ExChanGeAI simplifies the ECG deep learning workflow, making advanced analysis and model fine-tuning accessible to non-expert users with diverse datasets, while preserving privacy. The platform is available as open-source code under the MIT license.
