Application of Federated Learning in Chronic Wound Detection: A Distributed Approach to Patient Privacy Preservation

Diagnosing and managing chronic wounds presents considerable challenges in healthcare due to the complexity of tissue changes and the need for extended care. Recent advancements in artificial intelligence, particularly deep learning, have demonstrated significant potential for accurate analysis and automated diagnosis of medical images. However, concerns regarding the privacy and security of sensitive medical data—such as patient records and high-resolution imaging—pose substantial obstacles to the broader adoption of these technologies, particularly in distributed clinical environments. This study introduces Federated Learning (FL) as an innovative distributed approach for chronic wound diagnosis, enabling the training of AI models without the need to transfer sensitive patient data to centralized servers. By leveraging local model aggregation and distributed data processing across multiple healthcare institutions, this approach enhances diagnostic accuracy while ensuring robust protection of patient data. Unlike traditional centralized methods, FL preserves data locality, enabling scalable development of sophisticated diagnostic models within decentralized healthcare systems. The results demonstrate that FL-based models significantly outperform centralized approaches in terms of diagnostic accuracy while simultaneously reducing diagnostic errors. Additionally, integrating optimization algorithms and advanced security mechanisms strengthens the model's resistance to security threats, including cyberattacks targeting medical data. This research underscores the transformative potential of Federated Learning in advancing chronic wound diagnosis, with a strong emphasis on patient privacy in clinical settings.

کلیدواژه ها:

Artificial Intelligence ، Chronic Wound Classification ، Federated Learning ، Data Privacy