Pseudonymisation of Health Data: A Practical Assessment of Re-identification Risk