Conformity with AI Act – Annex IV
This page outlines how the Neolens documentation fulfills the requirements set forth in Annex IV of the EU Artificial Intelligence Act (AI Act) regarding the technical documentation for high-risk AI systems.
Each section of Annex IV is mapped below to its corresponding documentation area.
1. General Description of the AI System
Requirement:
A general description of the AI system, its intended purpose, and the system's overall logic.
Documentation Mapping:
- AI System Overview – Provides a high-level description of Neolens' purpose (medical image analysis), technical logic, and user base.
use-cases-by-industry.md– Lists specific real-world scenarios across clinical sectors.
2. System Development and Training
Requirement:
Information about the development process, training data sources, and data governance.
Documentation Mapping:
ai-concepts.md– Clarifies key terms related to model training and data handling.configuring-ai.md– Explains parameterization and configuration logic.ai-limitations.md– States known limitations, biases, and performance considerations.
3. Technical Parameters and Configurability
Requirement:
A detailed description of the system’s configuration, settings, and limits.
Documentation Mapping:
configuring-ai.md– Presents the types of configurable parameters and use cases.interpreting-results.md– Explains how parameter changes affect outputs.
4. Human Oversight Measures
Requirement:
Details of how human oversight is ensured during AI usage.
Documentation Mapping:
best-practices.md– Lists recommended human interventions and verification workflows.ai-troubleshooting.md– Supports diagnostic guidance for human operators.
5. System Performance and Accuracy
Requirement:
Information on accuracy, robustness, and cybersecurity resilience.
Documentation Mapping:
detection.mdclassification.mdai-limitations.md– Describes performance thresholds and known limitations.
6. Risk Management and Compliance
Requirement:
Details of the measures taken to comply with obligations under the AI Act.
Documentation Mapping:
gdpr.md– Describes GDPR alignment, data minimization, and security-by-design.ai-limitations.md– Covers identified ethical and legal constraints.
7. Record-Keeping and Traceability
Requirement:
Mechanisms for logging, traceability, and version control.
Documentation Mapping:
authentication.md– Details how API calls are authenticated and logged.api-troubleshooting.md– Explains how logs are used for error diagnosis.
8. Post-Market Monitoring
Requirement:
Plans and mechanisms to monitor the system after deployment.
Documentation Mapping:
- (Simulated) This demo documentation assumes a monitoring plan exists but is not implemented in this fictional example. Placeholder section may be added later.
Disclaimer
🧪 This documentation is part of a fictional portfolio project. It does not represent a real medical device or AI system, and is not legally certified under the AI Act. However, it is intentionally structured to simulate conformity with Annex IV of the EU AI Act to demonstrate good documentation practices.