Data De-identification

Data de-identification is the process of removing or transforming information that links data to a particular individual. It addresses both obvious identifiers, such as names, and combinations of attributes that may indirectly reveal identity. It reduces privacy risk, but does not by itself guarantee anonymity.

How de-identification works

Effective de-identification considers the dataset, other information reasonably available for linkage, and the circumstances in which the data will be used or disclosed.

• Direct identifiers, such as names or identification numbers, may be deleted, masked or replaced with tokens.
• Indirect identifiers, such as age, location and occupation, may be generalised, suppressed or aggregated because their combination can identify a person.
• Statistical techniques, including perturbation and controlled noise, can reduce the disclosure of information about particular individuals.

Pseudonymisation and anonymisation

De-identification covers techniques with different levels of protection, so the terms should not be treated as interchangeable.

• Pseudonymisation replaces identifying information with a code or pseudonym. If a key or additional information permits relinking, the data remains identifiable.
• Anonymisation seeks to make identification impracticable in the relevant context, including through combination with other datasets.
• The effectiveness of either approach depends on re-identification risk, technological change, data access and the uniqueness of retained attributes.

Role in digital privacy protection

Under the Digital Personal Data Protection Act, 2023, personal data means data about an individual who is identifiable by or in relation to such data. Therefore, transformed data remains personal data when an individual can still be identified from it, directly or through linkage.

• De-identification supports privacy by limiting unnecessary exposure, enabling safer analysis and reducing harm from unauthorised access.
• It should be combined with access controls, security safeguards, purpose limitation and periodic assessment of re-identification risk.
• Stronger de-identification generally protects privacy better but may reduce the accuracy or usefulness of data for research and analysis.