This paper proposes a formal model for expressing policies in digital health. The aim is to support computable expressions of legislative, regulative and organizational policies. The model is grounded in the semantics of deontic logic [1] and in modelling concepts for expressing accountability, specified in the new RM-ODP Enterprise Language standard [2]. An example of privacy consent based on the FHIR consent resource [3] is used to explain the use of these modelling concepts. The example involves multiple stakeholders and illustrates the complexity associated with the use of machine learning and artificial intelligence systems as part of healthcare delivery governed by informed consent policies.