Sistema de seguridad de red basado en el aprendizaje federado para detectar intrusiones en entornos médicos
Cargando...
Fecha
2026
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Católica Santo Toribio de Mogrovejo
Resumen
La ciberseguridad en entornos de salud, particularmente la protección de los dispositivos médicos conectados (DMC), presenta el desafío dual de mitigar el riesgo de intrusiones sin comprometer la privacidad del paciente. Esta investigación abordó dicha problemática mediante el desarrollo de un sistema de detección de intrusiones (IDS) basado en el aprendizaje federado. Se estableció una clasificación de amenazas jerárquica (Normal, Sospechosa y Ataque) tras analizar el tráfico de red del dataset “CICIDS2017”. El modelo Random Forest fue seleccionado como el algoritmo óptimo dentro de la arquitectura del aprendizaje federado, logrando una robustez de detección validada por métricas F1-Score y Recall. El sistema se implementó a través de un dashboard interactivo que permite la visualización y gestión en tiempo real del modelo federado. Finalmente, la usabilidad del sistema fue evaluada mediante el cuestionario SUS, alcanzando un puntaje promedio de 85.0, lo que lo clasifica con un nivel de usabilidad óptimo. Los resultados confirman que el sistema de seguridad de red basado en aprendizaje federado es una solución altamente precisa, eficiente y usable que aborda la necesidad de proteger los entornos médicos, garantizando la confidencialidad de los datos al entrenar los modelos localmente en los nodos de la red.
Cybersecurity in healthcare settings, particularly the protection of connected medical devices (CMDs), presents the dual challenge of mitigating the risk of intrusions without compromising patient privacy. This research addressed this issue by developing an intrusion detection system (IDS) based on federated learning. A hierarchical threat classification (Normal, Suspicious, and Attack) was established after analyzing the network traffic from the “CICIDS2017” dataset. The Random Forest model was selected as the optimal algorithm within the FA architecture, achieving detection robustness validated by F1-Score and Recall metrics. The system was implemented via an interactive dashboard that enables real-time visualization and management of the federated model. Finally, the system’s usability was evaluated using the SUS questionnaire, achieving an average score of 85.0, which classifies it as having an optimal level of usability. The results confirm that the federated learning-based network security system is a highly accurate, efficient, and usable solution that addresses the need to protect medical environments, ensuring data confidentiality by training models locally on network nodes.
Cybersecurity in healthcare settings, particularly the protection of connected medical devices (CMDs), presents the dual challenge of mitigating the risk of intrusions without compromising patient privacy. This research addressed this issue by developing an intrusion detection system (IDS) based on federated learning. A hierarchical threat classification (Normal, Suspicious, and Attack) was established after analyzing the network traffic from the “CICIDS2017” dataset. The Random Forest model was selected as the optimal algorithm within the FA architecture, achieving detection robustness validated by F1-Score and Recall metrics. The system was implemented via an interactive dashboard that enables real-time visualization and management of the federated model. Finally, the system’s usability was evaluated using the SUS questionnaire, achieving an average score of 85.0, which classifies it as having an optimal level of usability. The results confirm that the federated learning-based network security system is a highly accurate, efficient, and usable solution that addresses the need to protect medical environments, ensuring data confidentiality by training models locally on network nodes.
Descripción
Palabras clave
Ciberseguridad, Aprendizaje federado, Dispositivos médicos conectados, Cybersecurity, Federated learning, Medical IoT devices
Citación
J. Torres, “Sistema de seguridad de red basado en el aprendizaje federado para detectar intrusiones en entornos médicos,” tesis de licenciatura, Fac. de Ingeniería, Univ. USAT, Chiclayo, Perú, 2026. [En línea]. Disponible en: https://hdl.handle.net/20.500.12423/10262
