Using Health and Clinical Data to Train Algorithms and AI Models

JOTTERAND, Alexandre. Using Health and Clinical Data to Train Algorithms and AI Models: Legal Concerns and the Need (or Not) to Collect Consent. Life Science Recht, 3/2023, p. 130–146.

Abstract (AI generated)

Clinical and other health data are a goldmine for training AI algorithms and models, both in clinical settings and for commercial products. This article analyses the Swiss legal framework for such reuse: key data protection principles (lawfulness, legality, purpose limitation and secondary use), special rules under the Human Research Act and the Human Genetic Testing Act, limits on disclosure under data protection law, professional secrecy and international transfers, and additional obligations, with an excursus on secondary use under the GDPR and the European Health Data Space. Training AI models on patient data will generally fall under the Human Research Act and require ethics committee approval.