The health sector is steadily becoming the de facto target for cyberattacks. Based on the most recent ENISA report at the end of 2018, cybersecurity incidents have shown that the healthcare sector is one of the most vulnerable. Focusing specifically on Electronic Medical Devices (EMD), they suffer from numerous and multi-layered vulnerabilities . Default, weak or no password authentication for remote connections, unencrypted traffic or obsolete and insecure cryptographic algorithms, unsupported operating systems, outdated, unmanaged and vulnerable software are among the most serious problems that jeopardise both their smooth operation and the data aggregated and stored. The vision of HEIR is to provide a thorough threat identification and cybersecurity knowledge base system addressing both local (in the hospital / medical centre) and global (including different stakeholders) levels, that comprises the following pillars: (i) Real time threat hunting services, facilitated by advanced machine learning technologies, supporting the identification of the most common threats in electronic medical systems based on widely accepted methodologies such as the OWASP Top 10 Security Risks and the ENISA Top 15 Threats; (ii) Sensitive data trustworthiness sharing facilitated by the HEIR privacy aware framework; (iii) Innovative Benchmarking based on the calculation of the Risk Assessment of Medical Applications (RAMA) score, that will measure the security status of every medical device and provide thorough vulnerability assessment of hospitals and medical centres; (iv) The delivery of an Observatory for the Security of Electronic Medical Devices; an intelligent knowledge base accessible by different stakeholders, providing advanced visualisations for each threat identified in RAMA and facilitating global awareness on EMD-related threats. Last, HEIR will set up a broad European network for establishing good security practice in all regulatory frameworks to reduce market access.

Computerised Tomography (CT) scan is one of the most common medical imaging performed in healthcare, Each year, 300 million CT scans are performed globally. Of which, around 180M include use of radiocontrast media (RCM). Contrast Enhanced CTs (CECT) create a significant environmental impact, namely: 42,000 tonnes of single use packaging, 900 Tonnes of surgical steel (needles), 90,000 tonnes of plastic tubing and 150,000,000 kWh in energy consumption. These generate on average 9.2 kg of CO2/ scan. In addition, CECTs generates 200,000 tonnes of iodine contamination in water/yr. This is a recognised form of ‘pharmaceutical pollution’. CECTs may also harm patients: needle insertion, toxicity of iodinated RCMs to kidneys (potentially kidney failure) and allergic reactions, which in some cases can be life-threatening. Healthcare systems are responsible for the 4.4% CO2 global emissions (2 Giga tonnes/yr). Of this, ~3 Mega tonnes/yr are generated from CECTs. The EU has declared its NetZero targets of by 2050 through the European Green Deal. We showed feasibility that artificial intelligence (AI, deep learning methods) can extract high level information from non-contrast CT scans and synthesise contrast ‘digitally’. This avoids the need to administer RCM for CECTs. We seek to develop and validate 5 uses cases of CT ’Digital Contrast’ during this Horizon award. By implementing ‘Digital Contrast’ for scans globally, we aim to reduce 30% of the CO2e and iodine RCM waste generated from CECTs by 2033. NetZeroAICT has a grand vision to define a reference framework for scalable development of AI health tools for a future of sustainable health systems. This builds on our prior efforts of AICT consortium, which was established to make CT imaging safer, more efficient, more equitable and more sustainable. NetZeroAICT will accelerate the EU’s trajectory towards NetZero and advance EU’s globally recognized leadership position on Healthcare sustainability.

The data generated in the health domain is coming from heterogeneous, multi-modal, multi-lingual, dynamic and fast evolving medical technologies. Today we are found in a big health landscape characterized by large volume, versatility and velocity (3Vs) which has led to the evolution of the informatics in the big biodata domain. AEGLE project will build an innovative ICT solution addressing the whole data value chain for health based on: cloud computing enabling dynamic resource allocation, HPC infrastructures for computational acceleration and advanced visualization techniques. AEGLE will: - Realize a multiparametric platform using algorithms for analysing big biodata including features such as volume properties, communication metrics and bottlenecks, estimation of related computational resources needed, handling data versatility and managing velocity - Address the systemic health big bio-data in terms of the 3V multidimensional space, using analytics based on PCA techniques - Demonstrate AEGLE’s efficiency through the provision of aggregated services covering the 3V space of big bio-data. Specifically it will be evaluated in: a)big biostreams where the decision speed is critical and needs non-linear and multi-parametric estimators for clinical decision support within limited time, b)big-data from non-malignant diseases where the need for NGS and molecular data analytics requires the combination of cloud located resources, coupled with local demands for data and visualization, and finally c)big-data from chronic diseases including EHRs and medication, with needs for quantified estimates of important clinical parameters, semantics’ extraction and regulatory issues for integrated care - Bring together all related stakeholders, leading to integration with existing open databases, increasing the speed of AEGLE adaptation - Build a business ecosystem for the wider exploitation and targeting on cross-border production of custom multi-lingual solutions based on AEGLE.

The traditional cloud centric IoT has clear limitations, e.g. unreliable connectivity, privacy concerns, or high round-trip times. IntellIoT overcomes these challenges in order to enable NG IoT applications. IntellIoT’s objectives aim at developing a framework for intelligent IoT environments that execute semi-autonomous IoT applications, which evolve by keeping the human-in-the-loop as an integral part of the system. Such intelligent IoT environments enable a suite of novel use cases. IntellIoT focuses on: Agriculture, where a tractor is semi-autonomously operated in conjunction with drones. Healthcare, where patients are monitored by sensors to receive advice and interventions from virtual advisors. Manufacturing, where highly automated plants are shared by multiple tenants who utilize machinery from third-party vendors. In all cases a human expert plays a key role in controlling and teaching the AI-enabled systems. The following 3 key features of IntellIoT’s approach are highly relevant for the work programme as they address the call’s challenges: (1) Human-defined autonomy is established through distributed AI running on intelligent IoT devices under resource-constraints, while users teach and refine the AI via tactile interaction (with AR/VR). (2) De-centralised, semi-autonomous IoT applications are enabled by self-aware agents of a hypermedia-based multi-agent system, defining a novel architecture for the NG IoT. It copes with interoperability by relying on W3C WoT standards and enabling automatic resolution of incompatibility constraints. (3) An efficient, reliable computation & communication infrastructure is powered by 5G and dynamically manages and optimizes the usage of network and compute resources in a closed loop. Integrated security assurance mechanisms provide trust and DLTs are made accessible under resource constraints to enable smart contracts and show transparency of performed actions.

cPAID envisions researching, designing, and developing a cloud-based platform-agnostic defense framework for the holistic protection of AI applications and the overall AI operations of organizations against malicious actions and adversarial attacks. cPAID aims at tackling both poisoning and evasion adversarial attacks by combining AI-based defense methods (e.g., life-long semi-supervised reinforcement learning, transfer learning, feature reduction, adversarial training), security- and privacy-by-design, privacy-preserving, explainable AI (XAI), Generative AI, context-awareness as well as risk and vulnerability assessment and threat intelligence of AI systems. cPAID will identify guidelines to a) guarantee security- and privacy-by-design in the design and development of AI applications, b) thoroughly assess the robustness and resiliency of ML and DL algorithms against adversarial attacks, c) ensure that EU principles for AI ethics have been considered, and d) validate the performance of AI systems in real-life use case scenarios. The identified guidelines aspire to promote research toward developing certification schemes that will certify the robustness, security, privacy, and ethical excellence of AI applications and systems.