Healthcare

AI in healthcare can save lives, but also cause harm. Medical diagnostics, treatment recommendations and patient triage directly affect patient safety. The EU AI Act overlaps with the Medical Device Regulation (MDR) and IVDR. AI as a medical device is high-risk by definition. The stakes are highest here.

AI applications in this sector

Medical diagnostics

AI systems that analyze medical images (radiology, pathology, dermatology) or provide diagnostic suggestions. When certified as a medical device, these fall under Annex I, section A (via the MDR) and are high-risk by definition.

Treatment recommendations

AI systems that suggest treatment plans based on patient data, medical literature and clinical guidelines. The boundary between decision support and autonomous decision-making determines the risk classification.

Patient triage

AI for prioritizing patients based on urgency, symptoms and available capacity. Direct impact on patient safety when the system makes errors in urgency assessment.

Drug discovery

AI for identifying potential pharmaceuticals and predicting efficacy. During the research phase, this usually falls under minimal risk, but can become high-risk in clinical application.

AI as medical device

Software that qualifies as a medical device under the MDR (class IIa or higher). The EU AI Act explicitly references the MDR and IVDR in Annex I: these AI systems are automatically high-risk and must comply with both regulations.

High-risk classification

The EU AI Act (Regulation 2024/1689) classifies the following healthcare applications as high-risk:

AI as medical device

Annex I, section A (via MDR 2017/745)

AI systems that qualify as a medical device or in-vitro diagnostic medical device under the MDR or IVDR are automatically high-risk under the EU AI Act. This includes diagnostic software, clinical decision support systems and medical image analysis.

Safety components of medical devices

Annex I, section A

AI systems that function as a safety component of a medical device. This applies even when the AI system itself does not qualify as a medical device, but fulfills a safety-relevant function.

Access to healthcare

Annex III, point 5(a)

AI systems used by public authorities or on their behalf to evaluate whether persons are eligible for healthcare services. This affects triage, waiting list management and healthcare access decisions.

Specific challenges

Overlap with MDR and IVDR

The EU AI Act explicitly references the MDR (2017/745) and IVDR (2017/746). AI systems that qualify as medical devices must comply with both regulations. The conformity assessment procedures partially overlap but are not identical. You need expertise in both frameworks.

Patient safety

Errors in medical AI can directly lead to wrong diagnoses, missed conditions or incorrect treatments. The EU AI Act requires robust risk management (Article 9) and human oversight (Article 14). In healthcare, the threshold for "acceptable risk" is lower than in other sectors.

Clinical validation

Medical AI systems must be clinically validated. The MDR requires clinical evidence, the EU AI Act requires that training data is representative and of high quality (Article 10). Both requirements demand a structured validation approach.

Data sensitivity

Medical data is special category personal data under the GDPR (Article 9). The EU AI Act sets additional requirements for data governance of high-risk systems (Article 10). You need robust data processing agreements, pseudonymization and access controls.

Our approach for healthcare

Healthcare requires an approach that accounts for the overlap between the EU AI Act, the MDR/IVDR and the GDPR. We combine knowledge of AI regulation with understanding of the clinical context and help you set up an integrated compliance framework.

2 weeks

Compliance Quickscan

Inventory of all AI systems in your organization
Sector-specific risk classification per system
Gap analysis against EU AI Act requirements
Prioritized roadmap with concrete action items
Management presentation with findings and recommendations
1 day

AI Literacy Training (Article 4)

Sector-specific tailored training
Role-specific modules for your teams
Practice-oriented workshops with sector case studies
Proof of participation per employee
Reference materials and quick-reference cards
6 weeks

Governance Framework

AI policy aligned with your sector regulation
Roles and responsibilities (RACI matrix)
Risk management process for AI systems
Fundamental Rights Impact Assessment templates
AI registry with all required documentation
Monitoring and review cycle

In healthcare, AI compliance is about patient safety.

Medical AI affects people at their most vulnerable. The EU AI Act and the MDR rightly set the highest requirements here. In a free 30-minute intake we map out which AI systems your institution uses, how the MDR and EU AI Act apply to your situation, and what the first steps are.

Book your free intake

Not satisfied after the Quickscan? You pay nothing.

Rivium Westlaan 46, Capelle aan den IJsselCoC 90283597