Green Learning for Eco-Conscious Medical AI — a modular, energy-efficient, and privacy-preserving AI framework for healthcare. Integrating TinyML, federated learning, synthetic data generation, energy benchmarking, and post-quantum cryptography across 24 partners in 8 countries.
Medical AI consumes enormous resources and most projects never leave the pilot stage.
Training GPT-3 alone emits an estimated 500–600 tonnes of CO2. In healthcare, data fragmentation and strict privacy regulations prevent effective cross-institutional learning, and around 80% of healthcare AI projects fail to move beyond pilot stage.
Energy Cost
Medical AI training and deployment consumes enormous computational resources. Data centres supporting healthcare AI are growing faster than efficiency gains can offset, creating an unsustainable trajectory.
Data Silos
Patient data is fragmented across hospitals, clinics, and national systems. Strict privacy regulations (GDPR, HIPAA) prevent the centralised data pooling that conventional AI training requires.
Pilot Failure
Around 80% of healthcare AI projects fail to progress beyond the pilot stage. Integration complexity, interoperability gaps, and regulatory uncertainty create barriers to clinical deployment.
Security Threats
As quantum computing advances, current cryptographic protections for medical data will become vulnerable. Healthcare systems need post-quantum security before the threat materialises.
Six objectives for sustainable medical AI.
GLEAM targets a measurable reduction in AI energy consumption while maintaining or improving clinical accuracy, across multiple medical imaging and signal analysis modalities.
"GLEAM brings together clinical, industrial, and academic partners to prove that medical AI can be both effective and sustainable. Digital Tactics leads the integration work package that turns these research advances into deployable systems."
Energy-Efficient AI Architectures
Develop energy-efficient AI architectures for medical imaging and signal analysis that reduce computational overhead without sacrificing diagnostic accuracy.
Sustainability Benchmarking Toolkit
Create an open sustainability benchmarking toolkit (led by IMEC) to measure and compare the energy footprint of AI training and inference pipelines across different hardware and software configurations.
Federated Learning Infrastructure
Implement federated learning infrastructure across pilot sites, enabling cross-institutional model training without centralising patient data or violating privacy regulations.
Synthetic Health Data Generation
Integrate synthetic health data generation for regulation-compliant training, augmenting real datasets to improve model robustness while maintaining patient privacy.
Clinical Proof-of-Concept
Demonstrate clinical proof-of-concept use cases across multiple modalities, validating that green AI approaches deliver clinically acceptable performance in real-world settings.
>20% Energy Reduction
Target greater than 20% energy reduction versus conventional AI baselines, measured across training, inference, and deployment phases using the project's own benchmarking toolkit.
Core technologies powering sustainable medical AI.
GLEAM combines edge computing, privacy-preserving learning, and post-quantum security into an integrated framework for energy-efficient healthcare AI.
TinyML & Edge AI
Compact model architectures optimised for edge deployment, reducing data centre dependency and enabling AI inference directly on medical devices and local hardware.
Federated Learning
Distributed model training across hospital sites without centralising patient data. Models learn from multi-institutional datasets while data remains under local governance.
Synthetic Data Generation
Artificial health data that preserves statistical properties of real datasets without exposing patient information. Enables regulation-compliant model training and validation.
Post-Quantum Cryptography
Next-generation encryption protecting medical data against future quantum computing threats. Trust-as-a-Service layer secures federated learning communications and data exchanges.
Green AI Benchmarking Toolkit
Open-source measurement framework for quantifying AI energy consumption across training, inference, and deployment. Enables objective comparison of efficiency improvements.
Dynamic Cloud-Edge Orchestration
Intelligent workload distribution between cloud and edge resources, optimising for energy efficiency, latency, and data governance requirements in real-time.
Applying GreenCode to Medical AI Sustainability.
Digital Tactics is the sole UK partner in GLEAM, leading Work Package 5 (Integration, Deployment & Interoperability) and the UK consortium. Our contribution bridges software-level energy optimisation with the project's hardware and algorithmic efficiency work.
"Our GreenCode platform provides the software-level energy measurement and optimisation that complements IMEC's hardware benchmarking. Together, we can quantify the full stack energy impact of medical AI systems."
AI-Automated Code Optimisation
Applies GreenCode's AI-automated code optimisation to GLEAM framework components, identifying and eliminating energy waste at the software level across training pipelines, inference engines, and data processing workflows.
Software-Level Energy Savings
Benchmarks software-level energy savings targeting 10–30% reduction in computational energy consumption. Cross-validates results against IMEC's hardware-level benchmarks to provide full-stack efficiency measurement.
Sustainable Software in Regulated Domains
Provides best-practice guidelines for sustainable software engineering in regulated healthcare domains, addressing the unique constraints of medical device software, clinical validation requirements, and audit trails.
Joint GreenCode-GLEAM White Paper
Joint GreenCode-GLEAM white paper on sustainable medical software engineering, combining hardware benchmarking data with software optimisation results to establish a comprehensive efficiency methodology.
24 partners across 8 countries.
GLEAM brings together hospitals, research institutions, technology companies, and SMEs from across Europe and South Korea to develop and validate sustainable medical AI systems.
Related R&D Projects
View all R&D projects across Eureka, Horizon Europe, and internal research.
GreenCode
AI/ML Driven Software Optimisation to Reduce Cost and Climate Impact. Decarbonising computing at scale.
GIGI
Generative AI Inference Governance and Impact Measurement. Provider-agnostic AI usage control plane for enterprises.
Horizon CSA
Leveraging AI to optimise scientific computing for performance and energy efficiency. In development.
Interested in sustainable medical AI?
Get in touch to learn more about GLEAM and Digital Tactics' role in developing energy-efficient AI for healthcare.
AI software engineering and innovation
for ambitious organisations