Google Professional Machine Learning Engineer Question 129
Select 3Google Cloud PlatformYour company is building a machine learning model to predict patient outcomes based on electronic health records (EHR). The dataset contains protected health information (PHI), such as patient names, social security numbers, and medical histories. You are tasked with designing a data preprocessing pipeline on Google Cloud to ensure compliance with privacy regulations such as HIPAA. Which of the following actions should you take to handle PHI appropriately?
- A
Use Cloud Data Loss Prevention (DLP) API to detect and redact PHI in the dataset.
- B
Store the raw dataset containing PHI in a public Google Cloud Storage bucket for easy sharing within the team.
- C
Encrypt the dataset using Cloud Key Management Service (KMS) before storing it in Google Cloud Storage.
- D
Leverage differential privacy techniques to anonymize the dataset while preserving its utility for machine learning.
- E
Grant all members of your organization full access to the dataset to facilitate collaboration.
Show answer and explanation
Correct answers: A, C, D
Explanation
Handling PHI requires strict adherence to privacy regulations such as HIPAA. Using tools like Cloud DLP for redaction, encrypting data with Cloud KMS, and applying differential privacy techniques are industry best practices to ensure compliance. Avoid actions that expose sensitive data unnecessarily, such as storing it in public buckets or granting excessive access permissions.
- A. Correct.
Correct: Cloud Data Loss Prevention (DLP) API is designed to detect and redact sensitive information like PHI. This step is essential to ensure privacy compliance.
- B. Incorrect.
Incorrect: Storing raw PHI in a public bucket violates privacy regulations like HIPAA and can lead to severe security breaches.
- C. Correct.
Correct: Encrypting the dataset with Cloud KMS provides an additional layer of security for sensitive data, ensuring compliance with privacy standards.
- D. Correct.
Correct: Differential privacy techniques help anonymize PHI while maintaining the dataset's utility, making it a critical step for privacy-preserving machine learning.
- E. Incorrect.
Incorrect: Granting full access to sensitive data violates the principle of least privilege and privacy regulations. Access should be restricted to only those who need it.