Certification program for processing personal data in the context of prediction systems in health care

The present conformity assessment program was developed within the framework of the BMFTR-funded project “KI in der Pflege – Sturz, Delir, Medikation (KIP-SDM)” (AI in Nursing – Falls, Delirium, Medication). The primary aim of the project was to improve the quality of predicting fall events during treatment by using routinely collected patient or treatment data, and, in connection with this, to reduce the risk of fall events. The project examined how extensive clinical and nursing data sets can be used to train prediction models while at the same time protecting the interests and fundamental rights of patients and persons in need of care. Based on the findings of the project, this conformity assessment program was designed and developed. It serves as a preliminary stage for a data protection certificate pursuant to Article 42 GDPR and ensures the GDPR-compliant and thus legally safe application of the project results in nursing practice. The program covers the processing of personal data from healthcare and nursing institutions for the purpose of training prediction models, as well as the use of the trained models for the prediction and classification of risks and risk factors in a professional healthcare and nursing context. The program requires that the data be sufficiently anonymized and, where appropriate, synthetically generated before the models are trained. In addition, the program specifies the necessary protective measures, as well as the type and form of the evidence to be provided and the methods for verifying such evidence.