Table 1.

Shifts in the collection and use of health data and their implications for biomedical research

Shift	What is the challenge?
Changing concept of “personal data”	Personal data are personal for a wider range of people than the individual from which they were collected or otherwise processed.
Limits of anonymization	Data are never fully anonymized in the sense that the re-identification individuals becomes impossible. New opportunities for data linkage and the integration of different data sets can make re-identification possible. Further, anonymization may not be the best means to protect and promote the interests of both researchers and participants.
Added pressures on consent procedures	When data are collected and stored for future use, it is impossible to anticipate all future uses and thus require fully informed and specific consent.
Transferability of health data to other domains (and vice versa)	Virtually any data set can be used to make health-relevant inferences pertaining to individuals (especially in the context of predictive analytics). Thus, also data that were not collected for health-relevant purposes can be used in a health-relevant way.
Risks associated with predictive analytics	It is very difficult, if not impossible, for individuals to know what data are used to make inferences and predictions about them. If data are used to harm them, or if inaccurate data are used, there are typically few options to rectify the harm/error or seek redress.