In SurgicalInformatics/encryptr: Easily Encrypt and Decrypt Data Frame/Tibble Columns or Files using RSA Public/Private Keys

knitr::opts_chunk$set(echo = FALSE)

# Following line will delete any existing public and private keys in directory
# Do not uncomment it unless you are happy to delete these keys
# If the keys are present the code won't run as it attempts to create them (change the key names at gen_keys() to run slides if keys already exist and can't be deleted
# system(paste0("rm ", getwd(), "/id_rsa*"))
options(width=150)
options(dplyr.print_min = 6, dplyr.print_max = 6)

Data Governance {.fourlinestext}

• Data is precious and sensitive

• Confidential data (when possible) should be anonymised or pseudonomysed

• Patients expect us to safeguard their data

• Solutions include removing unnecessary data and securing necessary data

Research and Data {.fourlinestext}

• Data governance approvals are key components of research approvals

• GDPR (Europe) and HIPAA etc. (USA)

• Data breaches are financially and reputationally costly

• Not all data can be removed from records

Encryptr {.fourlinestext}

• Easy pseudonymisation by encryption

• RSA encryption with private / public key pair

• Encryption of vectors, variables and files

• Secure storage of confidential data (and allocation concealment / blinding)

Encryptr on CRAN / Github

install.packages("encryptr") # CRAN
remotes::install_github("SurgicalInformatics/encryptr")

https://github.com/surgicalinformatics

---

### <b>
library(encryptr)
### </b>
library(dplyr) # Used in presentation examples

# Encryptr comes with an example data set of GPs (Family Physicians)
### <b>
gp
### </b>

Public and Private Keys {.basictext}

### <b>
genkeys()
### </b>

Default values are "id_rsa" and "id_rsa.pub"

No Raw Text Password {data-background=#E1D9D7}

knitr::include_graphics("API_dialogue.png") 

---

### <b>
gp_encrypt = gp %>% 
  encrypt(name)
### </b>
gp_encrypt %>% 
  select(organisation_code, name, address1)

---

### <b>
gp_encrypt %>%
  slice(1:2) %>%
  decrypt(name) %>% 
  ### </b>
  select(organisation_code, name, address1)

Encryptr Customisation {.fourlinestext}

• Use look-up table - create object with encrypted output and ID variable on which to match

• Write a look-up file

• Customise file names, key names, encrypt several variables

• Use a publicly-available public key

Encryptr Customisation Examples

# Creating a lookup table with specified name and filename
### <b>
gp %>% 
  encrypt(name, postcode, 
          lookup = TRUE, write_lookup = TRUE, 
          lookup_name = "new_lookup_name")


# Using a public key hosted at URL
gp %>% 
  encrypt(name, public_key_path = "https://<some_url>/id_rsa.pub")
### </b>

Encryptr File Encryption

gp_encrypt %>% write_csv("gp_enc.csv")

### <b>
encrypt_file("gp.csv")



# Encrypted file will have suffix: `.encryptr.bin`
decrypt_file("gp.csv.encryptr.bin", file_name = "gp2.csv")
### </b>

Encryptr Ciphertexts Not Matchable {.fourlinestext}

• Each repeat of encryption generates a unique number

• Prevents malicious, opportunistic use of public key

• Alternative symmetric encryption outputs can be matched (and not always reversed)

• Alternative methods need a "salt"

---

### <b>
encrypt_vec(c("a name", "a name", "a name"))
### </b>

Technical Aspects of Encryptr {.fourlinestext}

• Wrapper around OpenSSL (and purrr)

• RSA asymmetric encryption for vectors (each component in vector < 256 bytes)

• File encryption uses AES technique with symmetric session key which is in turn encrypted by RSA public key

What Encryptr Is Useful For {.fourlinestext}

• Storing confidential data in trials, cohorts, service evaluation, etc.

• Retrieving individual data if needed

• Blinding in RCTs

• Data governance considerations

Questions

SurgicalInformatics/encryptr documentation built on Nov. 9, 2019, 6:55 p.m.