AI Genomic Data: Reidentification Risk

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-ethics-safety-ai-genomic-data-reidentification-risk-r10a4-adults

1. Why is 'anonymized' genomic data considered uniquely vulnerable to reidentification compared to other data types?

   1. Genomic data contains immutable biological patterns that can be matched against growing consumer databases
   2. Anonymization techniques are more advanced for genetic information than for other data
   3. Genomic data is automatically shared across all healthcare systems
   4. Genomic data is regulated by fewer privacy laws than financial records

2. When considering genomic data release, what specific metric should be estimated before proceeding?

   1. The number of SNPs in the dataset
   2. The computational cost of analysis
   3. Match probability against open genealogy databases
   4. The funding source for the research

3. Which of the following is a capability that AI systems can reliably provide regarding genomic data privacy?

   1. Guaranteeing that no individual can ever be reidentified from the data
   2. Ensuring complete compliance with all genetic privacy laws worldwide
   3. Running k-anonymity simulations to assess privacy thresholds
   4. Overriding IRB decisions about data release

4. Why must genomic data consent explicitly consider relatives of the data subject?

   1. Because one person's genome partially reveals the genetic makeup of their biological relatives
   2. Because relatives must physically accompany the subject to provide samples
   3. Because privacy laws only apply to family groups rather than individuals
   4. Because AI systems can only process data from multiple family members simultaneously

5. What does k-anonymity measure in the context of genomic data release?

   1. The total size of the genomic dataset in terabytes
   2. The duration of IRB review for a proposal
   3. The probability that an attacker will use machine learning
   4. The number of individuals who share identical genetic attributes in a dataset

6. What is the primary purpose of an IRB-ready risk memo in genomic data projects?

   1. To document the reidentification risks and mitigation strategies for ethical review
   2. To publish research findings in academic journals
   3. To calculate the budget for genetic sequencing equipment
   4. To recruit participants for clinical trials

7. What does GINA primarily protect against in the context of genetic information?

   1. Genetic data breaches caused by hacking attacks
   2. Unauthorized access to genetic sequencing equipment
   3. Discrimination by employers and health insurers based on genetic information
   4. International transfer of genomic data across borders

8. What makes consumer-genealogy databases particularly dangerous for genomic anonymity?

   1. They automatically delete data after 30 days
   2. They only accept data from accredited research institutions
   3. They contain millions of individuals' genetic profiles that can be used for matching
   4. They are regulated by international privacy treaties

9. When a researcher plans to release a genomic dataset, what should the consent process specifically address?

   1. The potential exposure of genetic information belonging to the subject's relatives
   2. The researcher's preferred publication venue
   3. The funding duration of the project
   4. The equipment used for DNA sequencing

10. Which task would be appropriate to assign to an AI system when preparing a genomic data release?

    1. Comparing different release strategies for privacy trade-offs
    2. Deciding whether to override IRB concerns about privacy
    3. Making the final decision on whether to release the data
    4. Determining whether the release complies with all international laws

11. What is the relationship between SNP sets and reidentification risk?

    1. SNPs are not used in genealogy matching algorithms
    2. Even small SNP sets can be sufficient to match individuals to genealogy databases
    3. SNPs reduce reidentification risk because they represent only a portion of DNA
    4. Only whole-genome sequences can be reidentified, not SNP sets

12. Why cannot AI systems override IRB judgment on genomic data releases?

    1. Because IRBs make ethical determinations that require human values and context that AI cannot replicate
    2. Because AI has already determined all genomic data is safe to release
    3. Because IRBs only review paper documents, not digital submissions
    4. Because AI systems lack the computational power to review protocols

13. What distinguishes reidentification risk in genomic data from reidentification in other datasets?

    1. Genomic data cannot be anonymized using any technique
    2. Genomic data is stored in specialized formats that are difficult to access
    3. Genomic data is smaller than other biomedical datasets
    4. Genomic data is immutable and can link to relatives, creating family-level exposure

14. If an AI system estimates a 15% match probability against a consumer genealogy database for a proposed genomic release, what should researchers conclude?

    1. The risk is significant enough to require additional mitigation before release
    2. The data can be released immediately without any concerns
    3. The AI system has malfunctioned and should be recalibrated
    4. The risk is negligible and IRB review is unnecessary

15. What are the fundamental limitations of AI in managing genomic privacy risks?

    1. AI cannot be used with genomic data at all
    2. AI cannot guarantee privacy and cannot replace human legal and ethical judgment
    3. AI cannot generate text outputs about genetics
    4. AI cannot process SNP data efficiently