Periodic Reporting for period 1 - ESCORT (AI ENABLED HEALTHCARE SERVICES DURING CROSS-BORDER MEDICAL EMERGENCIES AND REGULAR PATIENT SERVICES)
Reporting period: 2024-01-01 to 2025-06-30
The ESCORT project addresses one of the most pressing challenges facing European healthcare systems: ensuring resilience, efficiency, and continuity of care during crises while advancing the digital transformation of routine services. The COVID-19 pandemic, rising natural disasters, and increased cross-border mobility have exposed vulnerabilities such as unplanned patient surges, fragmentation between primary and social care, and weak coordination with emergency medical units. ESCORT responds by bringing together healthcare providers from six countries (Greece, Italy, Belgium, Ireland, Sweden, Israel), SMEs, industry, and research organisations to co-design and deploy digital health tools. Innovations include an AI-powered integrated healthcare system with smart handheld devices for emergency personnel and patients, telemedicine platforms for proactive interventions, and the integration of IoT, wearables, and edge devices to create interoperable, patient-centred services. The project’s pathway to impact combines evidence gathering (systematic reviews, user requirements, demand forecasting), technical innovation (AI algorithms, data integration, mobile apps, telemedicine), and pilot validation across six national contexts. These activities will generate actionable insights and deployable tools to strengthen resilience, improve surge management, and support cross-border services. Expected impacts are significant: reducing service disruptions during crises, improving patient experience and safety, optimising resources, and enabling long-term policy planning through AI-enabled modelling. ESCORT aligns with EU health priorities, including the European Health Data Space (EHDS) and Digital Europe Programme, while addressing the World Health Organization’s call for resilient, digitally empowered health systems. The integration of social sciences and humanities (SSH) is central: SSH experts contribute to co-design, ethics-by-design, and stakeholder consultations. Patient advocacy forums, workshops, and persona-driven design ensure ESCORT’s innovations remain inclusive, ethical, and aligned with human needs.
During the first reporting period, the ESCORT consortium advanced the project’s technical and scientific implementation, focusing on organisational requirements, system specifications, early prototypes, and validation methodologies. In WP2, a systematic review of over 140 studies on surge capacity planning (D2.1) provided the foundation for forecasting activities. Stakeholder engagement was strengthened through a patient advocacy forum (D2.2) embedding patient perspectives into design. These activities culminated in D2.3 – ESCORT User Requirements, v1, which mapped clinical and patient needs to the ten planned tools (R1–R10), capturing both scenario-specific and cross-cutting requirements.
In WP3 and WP6, partners jointly defined the ESCORT platform architecture, producing an integrated specification aligned with the three use cases. This, along with early prototypes, was consolidated in D6.3 – Pilot Demonstration and Validation (1st iteration), which reported preliminary tests, introduced the validation methodology, and outlined the initial integration and testing plan. In parallel, WP4 developed the basis for telemedicine services through patient personas and synthetic HL7 datasets, documented in D4.1 supporting mobile apps and visualisation tools for proactive interventions.
Complementary work in WP3 and WP5 advanced AI and semantic modelling. Algorithms were created to convert speech into structured HL7 data for automated emergency documentation, while a review of semantic models informed the interoperability strategy and alignment with HL7/FHIR and ISO/IEEE standards. Continuous integration and testing procedures were also established, enabling unified evaluation of tools against clinical and patient requirements. Scientific dissemination began with ESCORT’s first peer-reviewed paper, presented at ISCRAM 2025 and made publicly available under Open Science commitments.
In WP3 and WP6, partners jointly defined the ESCORT platform architecture, producing an integrated specification aligned with the three use cases. This, along with early prototypes, was consolidated in D6.3 – Pilot Demonstration and Validation (1st iteration), which reported preliminary tests, introduced the validation methodology, and outlined the initial integration and testing plan. In parallel, WP4 developed the basis for telemedicine services through patient personas and synthetic HL7 datasets, documented in D4.1 supporting mobile apps and visualisation tools for proactive interventions.
Complementary work in WP3 and WP5 advanced AI and semantic modelling. Algorithms were created to convert speech into structured HL7 data for automated emergency documentation, while a review of semantic models informed the interoperability strategy and alignment with HL7/FHIR and ISO/IEEE standards. Continuous integration and testing procedures were also established, enabling unified evaluation of tools against clinical and patient requirements. Scientific dissemination began with ESCORT’s first peer-reviewed paper, presented at ISCRAM 2025 and made publicly available under Open Science commitments.
The first iteration of pilot demonstrations reported in D6.3 confirms that ESCORT’s results (R1–R10) extend the state of the art in healthcare digitalisation and emergency preparedness. R1 (Resource management toolkit) advances beyond conventional static planning tools by dynamically optimising ICU capacity and triage decisions in real time during hospital surge scenarios. Preliminary validation during the hailstorm simulation at a virtual environment of St. George’s Hospital demonstrated its capacity to anticipate constraints and guide hospital resource allocation proactively. R2 (Integrated diagnostics and monitoring toolkit) and R3 (EMS–hospital integration via HL7/FHIR) surpass existing siloed solutions by creating an interoperable patient file across prehospital and hospital environments. R6 (AI self-reporting tool) and R7 (wearable-enabled monitoring) extend the state of the art in chronic disease management by embedding patient-generated data and wearable signals directly into interoperable EHRs. The early validation with personas (bronchiectasis, atrial fibrillation) demonstrated the feasibility of remote monitoring combined with telemedicine, reducing hospital readmissions while ensuring continuity of care. R8 (Predictive AI for treatments) introduces novel algorithms capable of assessing patient deterioration trajectories and optimising triage and ICU admission. This capability, absent in current decision support systems, enhances proactive care planning during both routine and surge conditions. R9 (Knowledge repository) and R10 (Pandemic forecasting) advance beyond existing public health surveillance systems by combining pathogen knowledge, AMR/CBRN threat data, and epidemiological modelling into a unified decision-support environment. A unifying advancement across all results is the interoperable data lake (IDL), which operationalises FAIR principles and enables multi-source data harmonisation in HL7/FHIR.