FRANCE Trends and Developments Contributed by: Liliana Eskenazi, Julie Ernewein and Pauline Lecrenais, Fréget Glaser et Associés
Ensuring risks are properly identified and addressed Finally, risk management is an essential component of any compliance strategy. High-risk healthcare AI sys - tems must be subject to a comprehensive risk man - agement process throughout their lifecycle. Devel - opers must identify and mitigate foreseeable risks to health, safety, or fundamental rights, implement appropriate safeguards, and report serious incidents. Healthcare institutions must also carry out clinical risk assessments prior to deployment, adopt technical and organisational safeguards, and provide staff with adequate training to ensure responsible interpretation of AI outputs. A robust compliance strategy must therefore inte - grate multiple layers, including transparency, regular internal audits, post-market monitoring, and adher - ence to data protection standards. On the user side, healthcare institutions and professionals are expected to adopt responsible practices aligned with legal and ethical standards, thereby ensuring the safe, effective, and trustworthy use of AI in medical care.
Key compliance requirements include differentiating anonymised vs pseudonymised data, ensuring data robustness for training/testing algorithms, and miti - gating bias and ensuring equitable algorithm perfor - mance Ensuring transparency and human oversight of health AI In line with European and national regulations, trans - parency and human oversight are also key require - ments for the development and use of high-risk AI systems. Providers must ensure that their systems are designed and documented in a way that is understandable to healthcare professionals and patients and must allow human intervention and control at all critical stages. Healthcare institutions and professionals, for their part, must apply AI tools ethically and transparently, retaining professional judgment (ie, not relying sole - ly on AI outputs) and informing patients when AI is involved in a preventive, diagnostic, or therapeutic act, in accordance with Article L. 4001-3 of the Pub - lic Health Code.
86
CHAMBERS.COM
Powered by FlippingBook