For projects handling personal or sensitive data, specific privacy techniques must be applied.

• De-identification: Removing all information that can help identify a person.

• Anonymisation vs. Pseudo-anonymisation: Anonymisation irreversibly removes identifiers. Pseudo-anonymisation replaces them with a reversible key, useful for clinical trials where follow-up data must be linked later.

• Differential Privacy & Synthetic Data: For high-risk data (e.g., medical DICOM images where AI can reverse-engineer identities from pixel matrices), apply Differential Privacy by injecting controlled statistical noise. Synthetic Data Generation (SDG) can mathematically preserve overall value distributions without exposing real individual records.