Research & Projects
Project portfolio, research areas, TRL levels and Open Science policy.
Research Projects
Each project follows the EU methodology: acronym, duration, budget, IRST role, Grant Agreement No.
AI-Driven Bio-Autonomous Micro-Pacemaker
Development of a micro-pacemaker powered by bioelectric energy harvesting from the human body, with integrated AI for real-time cardiac management.
Real-Time AI for Cardiac Rhythm Management
Development of artificial intelligence algorithms for real-time cardiac rhythm monitoring and management integrated into implantable devices.
Multi-Agent AI-Powered IDS for SME NIS2 Compliance on Ultra-Low-Resource Edge Devices
SENTINEL-AI is a multi-agent AI-powered Intrusion Detection System (IDS) designed for European SMEs that must comply with the NIS2 Directive (EU 2022/2555, Italian D.Lgs. 138/2024). The system operates on Raspberry Pi Zero 2 W hardware (<€25) with 8 specialised detection engines coordinated by 3 AI agents (VirusTotal, AbuseIPDB, Claude AI). In tests on a simulated network with 17 devices and 7 attack scenarios, it detected 15 threats (12 critical, 2 high, 1 medium) with zero false positives. It processes ~100 packets/second across 27 simultaneous devices, with ZeroClaw runtime (<5 MB RAM, <10 ms startup). Operating costs are two orders of magnitude lower than traditional MSSP solutions.
Cryptographic Certification Platform for Intellectual Property Protection
HashSigil is a cryptographic certification platform that provides proof of existence, integrity and authorship of any digital asset through triple-hashing (SHA-256, SHA-512, SHA-3-256), ECDSA signature, Bitcoin timestamping via OpenTimestamps and on-chain attestation on BNB Smart Chain via the proprietary A&T token. The system generates PDF certificates with evidentiary value under the eIDAS Regulation (EU 910/2014) and Italian Art. 2712 of the Civil Code, publicly verifiable through a unique URL and QR code. The serverless architecture is deployed on a global edge network with persistent PostgreSQL database, OTP email authentication and GDPR compliance (client-side hashing of all sensitive data).
Mutability Classification and Decoupled Cache System for AI Inference Optimisation
LiveInk implements the MC-TDC method (Mutability Classification – Three-tier Decoupled Cache), a three-level data mutability classification (static, slowly varying, volatile) that enables a decoupled caching strategy to reduce AI inference token consumption by up to 85%. The serverless architecture on Cloudflare Workers integrates over 15 external data sources (CoinGecko, ECB, Yahoo Finance, Numbeo, WAQI, USGS), Whisper AI audio transcription, multi-model fact-checking (GPT + Claude) and an embeddable JavaScript widget. The software is registered with the SIAE Special Public Register for Computer Programs (Reg. D000030735, 23/03/2026). Languages: TypeScript, JavaScript, Vue 3, HTML5, CSS3, SQL.
Secure Digital Asset Management System with Predictive AI and Cryptographic Certification
ATVault is an advanced secure digital asset management system developed by IRST, integrating advanced cryptographic protocols and predictive artificial intelligence. The security architecture operates on three levels: (1) SENTINEL-AI — 8 autonomous behavioural analysis agents scanning network, DNS, transactions, smart contracts and anomalies in real time; (2) HashSigil — dual-chain cryptographic certification (Binance Smart Chain + Bitcoin via OpenTimestamps) providing immutable public proof of each transaction; (3) end-to-end encrypted communication built into the wallet infrastructure. The system implements a meta-transaction relay with a native A&T token that fully abstracts gas fees from the user, and an on-chain digital inheritance protocol with time-based succession logic. AI intervenes autonomously at the pre-sign, simulation, recipient verification and contract analysis stages, eliminating user-side security complexity.
Technology stack: TypeScript, React, Solidity, Cloudflare Workers, Ethers.js, WebAuthn, EIP-712, OpenTimestamps, Meta-Transaction Relay
Departments and Expertise
Bioengineering and Medical Devices
Core Research AreaPrimary research area of IRST, dedicated to the development of autonomous medical devices powered by the human body.
- Bioelectric energy harvesting from the human body
- AI-driven autonomous micro-pacemakers
- Biocompatible and nanostructured materials
- Embedded systems for implantable devices
Medical Artificial Intelligence
Computational ResearchDevelopment of AI algorithms for medical applications, from predictive diagnostics to real-time management of implantable devices.
- Machine learning for cardiac diagnostics
- Edge AI for embedded devices
- Neural networks for bioelectric signals
- Digital twins for device simulation
Cybersecurity and Autonomous AI Systems
Security and DigitalisationDevelopment of autonomous AI agents for vulnerability analysis, critical infrastructure protection and secure digitalisation of assets on edge architectures.
Disclaimer: IRST's cybersecurity research activities are exclusively for defensive and civilian purposes. No technology developed falls within the dual-use categories under EU Reg. 2021/821.
- AI-driven autonomous penetration testing
- Security of IoT and implantable medical devices
- Edge computing and AI on resource-constrained hardware
- Smart contracts and blockchain technologies (verified on BscScan)
- Cryptographic certification and IP protection (HashSigil)
- Mutability classification and decoupled cache systems for AI optimisation (LiveInk)
- Secure digital asset management with predictive AI and dual-chain cryptographic certification (ATVault)
Open Science and Intellectual Property
Open Science Policy
IRST adheres to international Open Science standards. We guarantee Open Access to scientific publications and manage research data according to FAIR principles. Compliant with Plan S and Horizon Europe mandates.
Intellectual Property
We balance transparency with the rigorous protection of our bioengineering patents, following the principle: "as open as possible, as closed as necessary".
Research Integrity & COI
Policy compliant with the European Code of Conduct for Research Integrity (ALLEA 2023). Includes norms on authorship, plagiarism, data fabrication and conflict of interest disclosure.
Read the PolicyOpen Source & Reproducibility
Source code of published papers is available on GitHub under open-source licences. Each release automatically receives a DOI via Zenodo to ensure citability and reproducibility.
Ethics and Safety
Internal Ethics Committee
Ethical oversight of all research protocols, with particular attention to bioethics and human subject protection.
CE Approvals
CE certification pathway for medical devices in compliance with the MDR Regulation 2017/745.
FAIR Data Management
Research data management following FAIR principles, with a Data Management Plan for each funded project.