privacyR: Privacy-Preserving Data Anonymization

Description

Tools for anonymizing sensitive data in healthcare and research datasets. Helps protect patient privacy while keeping data useful for analysis.

Details

Main functions:

• anonymize_id - Anonymize patient identifiers

• anonymize_names - Anonymize patient names

• anonymize_dates - Anonymize dates (shift or round)

• anonymize_locations - Anonymize geographic locations

• anonymize_age - Anonymize ages into buckets

• anonymize_dataframe - Anonymize entire data frames

Disclaimer

While this package aids in anonymizing patient data, users must ensure compliance with all applicable regulations. The author is not liable for any issues arising from use of this package. See the DISCLAIMER file for complete terms.

Author(s)

Vikrant Dev Rathore

References

For more information on data anonymization best practices, see:

• HIPAA De-identification Guidance: https://www.hhs.gov/hipaa/for-professionals/privacy/special-topics/de-identification/index.html

• CDC Data Privacy: https://www.cdc.gov/phlp/php/resources/health-insurance-portability-and-accountability-act-of-1996-hipaa.html

• California DHCS List of HIPAA Identifiers: https://www.dhcs.ca.gov/dataandstats/data/Pages/ListofHIPAAIdentifiers.aspx

Anonymize Age by Buckets

Description

Groups ages into buckets for privacy protection. Default uses 10-year buckets (0-9, 10-19, etc.) which are useful for research. Ages 90+ are grouped together.

Usage

anonymize_age(x, method = c("10year", "hipaa"), custom_buckets = NULL)

Arguments

Value

A character vector of age buckets

Examples

ages <- c(25, 45, 67, 92, 15, 78)
anonymize_age(ages)  # Uses 10-year buckets by default
anonymize_age(ages, method = "hipaa")  # Use HIPAA buckets

Anonymize Patient Data in a Data Frame

Description

Main function to anonymize patient data in a data frame or data.table. Automatically detects and anonymizes columns based on data types and naming patterns, or you can manually specify columns. Different datasets get different anonymized values for better privacy.

Usage

anonymize_dataframe(
  data,
  id_cols = NULL,
  name_cols = NULL,
  date_cols = NULL,
  location_cols = NULL,
  age_cols = NULL,
  auto_detect = TRUE,
  detect_by_type = TRUE,
  date_method = "shift",
  date_granularity = "month",
  location_method = "generalize",
  age_method = "10year",
  use_uuid = TRUE,
  seed = NULL,
  dataset_specific = TRUE
)

Arguments

Value

A data frame with anonymized patient data (preserves data.table class if input was data.table)

Examples

# Basic usage with auto-detection
patient_data <- data.frame(
  patient_id = c("P001", "P002", "P003"),
  name = c("John Doe", "Jane Smith", "Bob Johnson"),
  dob = as.Date(c("1980-01-15", "1975-03-20", "1990-06-10")),
  location = c("New York, NY", "Los Angeles, CA", "Chicago, IL"),
  diagnosis = c("A", "B", "A")
)
anonymize_dataframe(patient_data, seed = 123)

# With month_year date granularity (YYYYMM format)
anonymize_dataframe(patient_data, date_method = "round", date_granularity = "month_year")

# Works with data.table
if (requireNamespace("data.table", quietly = TRUE)) {
  dt <- data.table::as.data.table(patient_data)
  anonymize_dataframe(dt)
}

# With UUID anonymization (default)
anonymize_dataframe(patient_data, seed = 123)

# Without UUID (sequential IDs)
anonymize_dataframe(patient_data, use_uuid = FALSE, seed = 123)

Anonymize Dates

Description

Anonymizes dates by shifting them by a random offset or rounding to a specified granularity. Shifting preserves relative time differences.

Usage

anonymize_dates(
  x,
  method = c("shift", "round"),
  days_shift = NULL,
  granularity = "month",
  seed = NULL
)

Arguments

Value

A Date vector of anonymized dates (or character vector for "month_year" granularity)

Examples

dates <- as.Date(c("2020-01-15", "2020-03-20", "2020-06-10"))
anonymize_dates(dates, method = "shift", seed = 123)
anonymize_dates(dates, method = "round", granularity = "month")
anonymize_dates(dates, method = "round", granularity = "month_year")

Anonymize Patient Identifiers

Description

Replaces patient identifiers with anonymized versions while maintaining referential integrity (same IDs get the same anonymized value).

Usage

anonymize_id(x, prefix = "ID", seed = NULL, use_uuid = TRUE)

Arguments

Value

A character vector of anonymized identifiers

Examples

ids <- c("P001", "P002", "P003", "P001")
anonymize_id(ids)
anonymize_id(ids, prefix = "PAT", seed = 123)
anonymize_id(ids, use_uuid = FALSE, seed = 123)  # Use sequential IDs

Anonymize Geographic Locations

Description

Anonymizes geographic locations by removing them or replacing with generic labels. Maintains referential integrity (same locations get the same value).

Usage

anonymize_locations(
  x,
  method = c("remove", "generalize"),
  prefix = "Location",
  seed = NULL,
  use_uuid = TRUE
)

Arguments

Value

A character vector of anonymized locations

Examples

locations <- c("New York, NY", "Los Angeles, CA", "Chicago, IL")
anonymize_locations(locations, method = "remove")
anonymize_locations(locations, method = "generalize", seed = 123)
anonymize_locations(locations, method = "generalize", 
                    use_uuid = FALSE, seed = 123)  # Use sequential IDs

Anonymize Patient Names

Description

Replaces patient names with anonymized identifiers while maintaining referential integrity (same names get the same anonymized value).

Usage

anonymize_names(x, prefix = "Patient", seed = NULL, use_uuid = TRUE)

Arguments

Value

A character vector of anonymized names

Examples

names <- c("John Doe", "Jane Smith", "Bob Johnson")
anonymize_names(names)
anonymize_names(names, prefix = "PAT", seed = 123)
anonymize_names(names, use_uuid = FALSE, seed = 123)  # Use sequential IDs

Generate Dataset-Specific Seed

Description

Internal function to generate a seed based on dataset content. This ensures different datasets get different anonymized values even with the same user-provided seed.

Usage

generate_dataset_seed(data, user_seed = NULL)

Arguments

Value

A numeric seed value

Generate Short UUID for Anonymization

Description

Internal function to generate short, reproducible UUIDs for anonymization. Uses a hash-based approach to ensure referential integrity (same input always produces same UUID) while maintaining uniqueness across datasets.

Usage

generate_short_uuid(x, prefix = NULL, seed = NULL, length = 8)

Arguments

Value

Character vector of short UUIDs