Pseudo-Anonymised ID Generation

This interface enables the use of the SHA-1 hashing algorithm on CSV files for the generation of pseudo-anonymised IDs, currently used for the AIMBraTS project. The application operates entirely client-side, ensuring that no data is transmitted outside of the local machine. To begin, simply go to: Pseudo-Anonymised ID Generator

Motivation

For the entirety of the AIMBraTS project, we aim to have a pseudo-anonymised ID, that is unique to each person, but also allows us to cross reference this ID with the original patient information, if required.

The way we have decided to do this is by using a hashing algorithm, specifically SHA-1, to hash the NHS number. This hashed value is then truncated to only include the first 10 values of the generated hash, for readability.

The key advantage of this approach to generating pseudo-anonymised IDs is that with the ID, it is impossible to recreate the original input, and sensitive, data. But if someone has access to the original data, then they can use that to cross reference which patient has which pseudo-anonymised ID, thus providing a failsafe.

How to Use the Application

Step 0: Preliminary Data Check

Ensure each NHS number in your dataset is 10 digits long without any spaces. Errors in this data will lead to incorrect hashing.

Step 1: Access the Application

Navigate to the Pseudo-Anonymised ID Generator.

Note: This HTML application, which is accessed and viewed via a web browser, runs locally on your own machine, and no data is transmitted outside of the local machine, when used.

Step 2: Load Your Spreadsheet

The application currently supports .csv files only. Make sure your data is saved in this format before proceeding. Select your file by clicking on Choose File.

Step 3: Configure Data Columns

Assign roles to each column in your dataset, specifying which should be hashed, excluded, or kept. Specifically, assign the NHS number column to Hash and Exclude and all identifiable information to Exclude. Relevant clinical information, but not patient identifiable, can be assigned to Keep. Any junk that you do not want, in the anonymised spreadsheet, can be set to Exclude. Columns that you want a hashed version of, but you also want to appear in the anonymised spreadsheet can be set to Hash

Step 4: Process the Data

Click Process to start the local hashing. This generates two .csv files:

• original_with_hash.csv: Contains all original data along with the generated hashes.
• unidentifiable.csv: Contains only non-identifiable data and hashes.

Data Handling

• Internal Use: Keep original_with_hash.csv within the NHS trust.
• External Sharing: unidentifiable.csv is the anonymised spreadsheet and can be shared with researchers at KCL, following approved ethics and data sharing agreements.

Support and Contact

For any inquiries or support requests, please contact:

• Theodore Barfoot at [email protected]
• OR: [email protected]

Example Python Implementation

This operation performed by this application is equivalent to the following:

def hash_algo(nhs_number):
    return hashlib.sha1(nhs_number.encode()).hexdigest()[:10]