Responsible AI
This page integrates and discloses the AI governance and accountability framework of i-SENS aligned with the EU AI Act (effective 2024), high-risk medical device AI, IEEE 7000 ethical design, OECD AI Principles, and the UNESCO Recommendation on the Ethics of AI. The measurement algorithm of the Continuous Glucose Monitoring (CGM) device CareSens Air is classified as a High-Risk AI System that directly affects patients, and is subject to strict governance in accordance with global standards.
1. Key KPI Summary (2025)
Area | Key KPI | 2025 Performance |
AI Incidents and Malfunctions | CGM algorithm measurement error rate | ISO 15197 compliance |
AI Governance | High-risk AI registration and clinical evaluation | EU CE MDR + AI Act simultaneously satisfied |
AI Post-Market Surveillance | Malfunction notifications under EU MDR Article 87 | 0 cases |
Learning Data Governance | De-identification of patient identification information | 100% |
2. AI Ethics Policy — Five Principles
i-SENS operates five AI ethics principles integrating the OECD AI Principles (2019) + UNESCO Recommendation on the Ethics of AI (2021) + EU AI Act (2024).
· Human-Centric: AI assists in patient health decisions, while the final medical judgment is the responsibility of the patient and the healthcare professional. AI limitations are clearly indicated.
· Safety: Integrated application of ISO 15197 accuracy criteria + EU MDR clinical evaluation + ISO 14971 risk management.
· Transparency: Disclosure in the instructions for use of the operating principles, limitations, and accuracy of the measurement algorithm.
· Fairness and Non-Discrimination: Inspection of demographic bias in learning data (age, gender, race, body type).
· Accountability: AI Governance Committee + Person Responsible for Regulatory Compliance (PRRC) + incident response framework.
3. EU AI Act Alignment (Classification as a High-Risk AI System)
The EU AI Act classifies AI systems into four risk grades. The i-SENS CGM CareSens Air falls under high-risk AI, and meets the following 8 requirements.
No. | EU AI Act Requirement | i-SENS Implementation |
1 | Risk management system (Article 9) | Integrated application of ISO 14971 medical device risk management |
2 | Data and data governance (Article 10) | Regular inspection of the source, accuracy, and bias of learning data + 100% de-identification of patients |
3 | Technical documentation (Article 11) | EU MDR CER + Technical Documentation + Software Lifecycle (IEC 62304) |
4 | Record-keeping (Article 12) | Regular retention of measurement logs and error records (medical device traceability standards) |
5 | Transparency and provision of information (Article 13) | Specification of the operating principles, limitations, and accuracy of AI in the instructions for use (such as the recommendation to use finger-prick blood sampling in the event of rapid blood glucose fluctuations) |
6 | Human oversight (Article 14) | Healthcare professionals and patients retain final decision-making authority — AI is an auxiliary tool |
7 | Accuracy, robustness, and security (Article 15) | Compliance with ISO 15197:2013 accuracy ±15% 95% + compliance with EMC IEC 60601-1-2 |
8 | Quality management system (Article 17) | ISO 13485 QMS + review of introduction of ISO/IEC 42001 AI Management System [TBD] |
4. CGM Algorithm Governance
The CareSens Air Continuous Glucose Monitoring (CGM) device operates through four stages: enzymatic reaction + electrochemical signal + algorithm calibration + wireless transmission; the governance of each stage is as follows.
Stage | AI/Algorithm Role | Governance |
1. Enzymatic Reaction | Blood glucose oxidation — generation of electrochemical signal | ISO 15197 accuracy criteria |
2. Electrochemical Signal | Measurement of current in μA units | IEC 60601-1-2 EMC |
3. Algorithm Calibration | Correction for temperature, interference, and body fluid differences | Clinical data learning + EU CER certification |
4. Wireless Transmission | Transmission to smartphone and monitor | ISO 27001 information security + GDPR · HIPAA |