Leader: Rocco Oliveto (UNIMOL); Other collaborator(s): Emanuela Guglielmi (UNIMOL), Franz Iannucci (UNIMOL), Alessio Maselli (UNIMOL), Davide Donato Russo (UNIMOL), Pasquale Trinchese (UNIMOL)
This task aims to establish an appropriate e-health platform and to design and progressively enhance an advanced decision support system to support medical staff in the identification of abnormalities in older adults. The platform is conceived as device-agnostic, enabling the integration of heterogeneous wearable devices and ensuring flexibility and scalability across different clinical and home-care scenarios. The activities focus on the monitoring of cardiovascular conditions, leveraging data acquired from wearable devices for cardiac monitoring. The system supports the identification of abnormalities through advanced data analysis techniques, with particular attention to the interpretability of the results to effectively assist medical staff in clinical decision-making.
During this period, the activities focused on the design of the system and the definition of its architecture. In particular, the work included the analysis and specification of clinical and functional requirements, the definition of the main user roles and interaction workflows, and the characterization of data acquisition processes from wearable devices, with a specific focus on cardiovascular monitoring.
Based on these activities, a device-agnostic architectural solution was designed, identifying the main system components and their interactions, including data acquisition, storage, and processing modules. This phase resulted in a coherent architectural framework, independent of specific devices, to be instantiated in subsequent phases.
During this period, the activities focused on the instantiation and validation of the previously defined device-agnostic architecture. In particular, the work included the selection of a set of wearable devices for cardiovascular monitoring, produced by AEBiosystem (EasyTemp, EasyRing, and EasyQ), following a thorough analysis conducted by clinical partners. These devices enable the acquisition of key vital parameters, including body temperature, heart rate, oxygen saturation, ECG, and thoracic impedance.
Their integration allowed the definition and verification of the data acquisition workflow, in which devices transmit data through a home router to a remote server, whose data are then made available for integration into the platform. This phase confirmed the feasibility of the proposed architecture and demonstrated its capability to effectively support the integration of real devices, while preserving its device-agnostic nature and openness to alternative solutions.
During this period, the activities focused on the consolidation of the technological baseline and on the refinement of the system architecture. In particular, the RESCO telemedicine platform was acquired and adopted as a reference solution, enabling a more structured implementation of the previously defined architectural framework. Building on this baseline, a re-engineering process was initiated, with the objective of aligning the platform with the device-agnostic architecture defined in the previous phases. The activities included the analysis of the existing components, the identification of required adaptations, and the redesign of key modules to support scalable data acquisition, integration, and management.
In this context, a first prototype of the platform was developed and demonstrated in Termoli on September 11–13, 2024, within the Spoke 5 event “Sustainability of care systems for the elderly in an ageing society”. This demonstration provided an initial validation of the proposed solution in a real dissemination context. This phase marked the transition from architectural validation to concrete system implementation, establishing a solid foundation for the subsequent development and release of the platform.
During this period, the activities focused on the development and initial release of the platform. Building upon the re-engineering efforts carried out in the previous phase, the core components of the system were implemented, enabling end-to-end data acquisition, storage, and visualization. In particular, the platform was extended to support the management of users and monitoring workflows, as well as the integration of data collected from wearable devices.
Basic functionalities for the visualization and inspection of physiological parameters were also introduced, allowing healthcare professionals to access and monitor patient data within the system. This phase resulted in a first operational version of the platform, providing a working environment to support subsequent experimentation and further system evolution.
During this period, the activities focused on the enhancement of the platform capabilities and on the preparation of advanced data analysis functionalities. Efforts were devoted to the testing and consolidation of communication with the selected wearable devices, as well as to the improvement of the data acquisition pipeline.
At the same time, preliminary activities aimed at enabling automatic anomaly detection were initiated, with the objective of supporting the identification of relevant patterns in physiological data. In parallel, a native integration strategy between the platform and the devices was explored, with the objective of reducing dependency on intermediary components and improving system efficiency and robustness.
This phase prepared the ground for the subsequent integration of Artificial Intelligence techniques within the platform.
During this period, the activities focused on the consolidation of the platform and on the advancement of data analysis functionalities. In particular, the data acquisition and processing pipeline was further refined, improving system stability and reliability in view of experimental activities.
In parallel, research and experimentation activities on Machine Learning techniques for the analysis of physiological data were carried out, with particular attention to anomaly detection on cardiovascular signals, and specifically on ECG data. These activities contributed to the identification of suitable models and approaches for integration within the platform.
This phase strengthened both the technological and analytical components of the system, enabling the transition towards a more integrated and intelligent monitoring solution.
During this period, the activities focused on the further evolution of the platform and on the integration of advanced functionalities. In particular, the system was extended with Artificial Intelligence components for the analysis of ECG signals and the identification of anomalies related to cardiovascular conditions.
In parallel, the design of a conversational system was carried out, with the objective of enabling natural language interaction with the entire clinical record managed within the CARES platform. The conversational layer was conceived to support medical staff in accessing, exploring, and interpreting clinical information, including patient data, monitoring results, and clinical documentation.
The chatbot is designed as a Large Language Model-based system tailored to the medical domain, capable of understanding and answering clinical queries in a contextual and accurate manner, interpreting health data in real time, and supporting a natural navigation of the clinical record, also through voice-based interaction. The design activities included the analysis of the existing platform architecture, the identification of relevant clinical use cases, and the definition of domain-specific conversational workflows aligned with medical terminology, as well as requirements related to security, privacy, and integration within clinical workflows.
This phase represents a key milestone towards the release of the next version of the platform, scheduled for delivery by 31/12/2025.
Research articles:
Presentations
Software