THE VR SCHOOL
  • Schools & Districts
    • License the academy
    • Free program for your students
    • Request a demo
  • Catalog
    • AI & Data
    • Cybersecurity & Web
    • Programming
    • Health Science
    • All Guilds
    • All courses
  • Families
  • Accreditation
  • Log In
  • Request a demo
  • Home
  • Courses

RLP AI Design A — Machine Learning, Media and the Human Body

Curriculum

  • 4 Sections
  • 20 Lessons
  • Lifetime
Expand all sectionsCollapse all sections
  • Unit 1: How Machines Learn: Making AI Visible
    5
    • 1.1
      Explaining the Invisible: Designers Who Made Data Visible
      50 mins
    • 1.2
      Sorting by Features: Training a Classifier by Hand
      100 mins
    • 1.3
      Critiquing AI Explainers: Accuracy, Clarity and Hype
      50 mins
    • 1.4
      Designing the Explainer: Hierarchy, Sequence and Storyboard
      100 mins
    • 1.5
      Portfolio Task — How a Machine Learns: Explainer and Storyboard
      200 mins
  • Unit 2: Human-Centered AI: History, Ethics and Authorship
    5
    • 2.1
      Drawing Time: Timeline Design and the History of AI
      50 mins
    • 2.2
      Verified Event Cards: Building a Physical Timeline Storyboard
      100 mins
    • 2.3
      Who Is Responsible? Bias and Human-Centered AI in Medicine
      100 mins
    • 2.4
      Authorship and AI Imagery: Making the Presentation Yours
      150 mins
    • 2.5
      Portfolio Task — The Evolution of a Healthcare AI Innovation
      200 mins
  • Unit 3: Drawing the Body: The Eleven Organ Systems
    5
    • 3.1
      Art and Anatomy: Five Centuries of Seeing Inside the Body
      50 mins
    • 3.2
      Sculpting a System: Modeling Structure and Function
      100 mins
    • 3.3
      Clinician and Patient: Role-Play and the Critique of Health Communication
      100 mins
    • 3.4
      Illustrating a Condition: Panels for One Body System
      150 mins
    • 3.5
      Portfolio Task — The Body Atlas Exhibition
      200 mins
  • Unit 4: Signals and Diagnosis: Visualizing Vital Signs and Medical Data
    5
    • 4.1
      Data Portraits: Du Bois, Einthoven and the Art of the Signal
      50 mins
    • 4.2
      Pulse to Pixels: Coding a Heart-Rate Visualization
      100 mins
    • 4.3
      Who Sees Your Data? Medical Records, Genetics and HIPAA
      50 mins
    • 4.4
      Imaging the Invisible: Designing a Patient Brochure
      150 mins
    • 4.5
      Portfolio Task — Reading the Body’s Signals
      200 mins

Who Sees Your Data? Medical Records, Genetics and HIPAA

Unit 4  ·  Critique & Discourse Lab  ·  Lesson 18 of 20

Who Sees Your Data? Medical Records, Genetics and HIPAA

MA:Re9A.IIRLP-AIDA
Abstract medical data forms sit behind a lock while dots flow toward a family silhouette.
By the end of this lesson I can…

distinguish the main kinds of medical data and explain the basic protections of the HIPAA Privacy Rule, and evaluate health data visualizations for what they reveal about people.

Instruction

Medical AI runs on data about people. Before you design with health data, you need to know what kinds exist, who is allowed to see them and how a visualization can reveal more than it should.

Kinds of medical data. Clinical records include diagnoses, medications, allergies, test results, vital signs and clinicians’ notes, now mostly kept in electronic health records. Imaging data includes X-rays, CT and MRI scans and ultrasound, stored as digital images with information about the patient attached. Genetic data comes from tests that read parts of a person’s DNA; it is unusual because it also says something about their biological relatives. Device and wearable data includes readings from home monitors and fitness trackers. Clinicians combine these: an image suggests a problem, lab results and history narrow it down, and genetic information may guide treatment.

HIPAA. The Health Insurance Portability and Accountability Act of 1996 led to the HIPAA Privacy Rule, which HHS enforces. According to the HHS summary, it applies to covered entities (health plans, health care clearinghouses and health care providers who conduct certain transactions electronically) and their business associates. It protects individually identifiable health information, called protected health information. It limits how that information can be used and shared, and gives patients rights, including the right to see and get a copy of their records. It does not cover everyone who holds health data: many consumer apps and wearables are outside HIPAA unless they work on behalf of a covered entity. Other laws add protection. The federal Genetic Information Nondiscrimination Act (2008) restricts the use of genetic information by health insurers and employers. States can add their own rules: Missouri law, for example, says a health insurer may not require or request a genetic test, and may not consider a person’s genetic test results without that person’s approval, when deciding on health-plan coverage or premiums (Revised Statutes of Missouri, section 375.1303).

De-identification. Researchers and AI developers often use data with identifiers removed. HIPAA’s Safe Harbor method lists 18 kinds of identifiers to remove, such as names, most dates and small geographic units. But combining datasets can sometimes re-identify people, especially when a group is small or a person’s condition is rare.

How visualizations reveal people. A map of cases by neighborhood with only two cases in one block can point to a household. A chart of a small club’s health survey by gender and age may single out one member. A screenshot of a patient record in a presentation shows a name in the corner. Designers protect privacy by aggregating (grouping into larger categories), suppressing small counts, and using synthetic or illustrative data for demonstrations.

This lab. You will critique two health data graphics using the unit criteria plus a privacy criterion: could a viewer learn something about an identifiable person? Then write a structured debate on a dilemma: a hospital wants to share de-identified scans with a company training an AI tool that could help many patients. What should patients be told, and should they be able to say no? You argue both sides yourself: give each side its strongest case, restate each in a form its holder would accept before answering it, and only then add your own position. If you like, a family member can read the opposing side aloud while you answer.

📖 Read for this lesson

The reading opens over the lesson; close it to come back to exactly where you were. The checks below draw on it.

  • Health Data and Privacy: HIPAA in the Regulations’ Words, Missouri Genetic-Testing Law, Two Dashboards to Critique and the Scan Dilemma as a Written Debate45 CFR 160.103 and 164.514 and 164.524 (eCFR, retrieved Sept. 26, 2026); HHS OCR, Health App Use Scenarios and FTC Health Breach Notification Rule guidance (paraphrased); NHGRI, Genetic Discrimination (updated Jan. 6, 2022); Revised Statutes of Missouri section 375.1303 (revisor.mo.gov, paraphrased, retrieved Nov. 10, 2026); SLS original dashboards, dilemma and positions · about 9 min

    The HIPAA definitions and the 18 identifiers in the regulations’ own words, GINA and Missouri’s genetic-testing law for health insurers, two dashboards described closely enough to critique with a model critique, the privacy tests with arithmetic, and the scan dilemma with two example positions and a frame for a written debate in which you argue both sides.

Vocabulary in context

  • protected health information (PHI) — Individually identifiable health information held by a HIPAA covered entity or business associate.
  • covered entity — Under HIPAA, a health plan, health care clearinghouse or health care provider that conducts certain transactions electronically.
  • de-identification — Removing or changing information so that data can no longer reasonably identify a person.
  • aggregation — Combining individual records into larger groups so that no single person stands out.

Formative check

Work through these before moving on. They are not graded — they tell you whether the standard below has landed yet.

+50 XP

A free fitness app on your phone records your heart rate. It does not work on behalf of any health provider or plan. Does the HIPAA Privacy Rule cover that data?

HIPAA applies to covered entities and business associates. Many consumer apps fall outside it.
Myth or Fact?

A chart of a small club’s health survey broken down by gender, age and condition can reveal information about one identifiable member even with no names on it.

Small groups can single people out; aggregate or suppress small counts.
Flashcard

Clinical records

Tap to reveal
Answer

Diagnoses, medications, tests, notes, vital signs

Flashcard

Imaging data

Tap to reveal
Answer

X-ray, CT, MRI, ultrasound images with attached patient information

Flashcard

Genetic data

Tap to reveal
Answer

DNA test results that also say something about relatives

Flashcard

GINA (2008)

Tap to reveal
Answer

Limits use of genetic information by health insurers and employers

Write the structured debate on the de-identified scans dilemma: (1) Position A’s strongest case; (2) Position B’s strongest case; (3) a restatement of each position in a form its holder would accept, with a reply to it; (4) your own position on what patients should be told and whether they should be able to say no, with one reason that uses a fact from the reading.

0 words
Quick self-check

How confident are you that you can explain the kinds of medical data and the basics of HIPAA, and critique data graphics for privacy risks?

Not yetVery confident

Practice

Work these on paper or in your notebook, then open Check your answer. Aim for all of Fluency and Application; try at least one Challenge.

Printable version: this unit’s practice workbook (PDF)

Fluency

Build speed and accuracy with the core skill.

  1. Name the four kinds of medical data the lesson describes, with one example of each.
    Check your answer
    Answer: Clinical records (for example diagnoses, medications, test results, notes); imaging data (X-ray, CT, MRI, ultrasound); genetic data (DNA test results); device and wearable data (home monitors, fitness trackers).
    Genetic data is unusual because it also says something about a person’s biological relatives.
  2. Define covered entity, protected health information and aggregation.
    Check your answer
    Answer: Covered entity: a health plan, health care clearinghouse or health care provider that conducts certain transactions electronically. Protected health information: individually identifiable health information held by a covered entity or business associate. Aggregation: combining records into larger groups so no single person stands out.
    Business associates are also covered by the HIPAA Privacy Rule.
  3. What does GINA (2008) restrict, and what does Missouri section 375.1303 add?
    Check your answer
    Answer: GINA restricts the use of genetic information by health insurers and employers. Missouri law says a health insurer may not require or request a genetic test, and may not consider a person’s genetic test results without that person’s approval, when deciding on coverage or premiums.
    States can add their own rules on top of federal law.

Application

Use the skill in context. Show your reasoning.

  1. Does the HIPAA Privacy Rule generally cover each of these? (a) a hospital’s electronic health record; (b) a free fitness tracker app that works for no health provider or plan; (c) a company processing records on a hospital’s behalf as a business associate.
    Check your answer
    Answer: (a) Yes, a provider is a covered entity. (b) Generally no; many consumer apps are outside HIPAA. (c) Yes, business associates are covered.
    HIPAA applies to covered entities and their business associates, not everyone who holds health data.
  2. An invented neighborhood case map shows 140 cases in one district and 2 cases in one city block. Identify the privacy risk and give two fixes from the lesson.
    Check your answer
    Answer: The block with 2 cases could point to a household. Fixes: aggregate into larger areas (for example, show the whole district), or suppress small counts.
    Designers also use synthetic or illustrative data for demonstrations.
  3. A presentation includes a screenshot of a real patient record with the name visible in the corner. Using the lesson, explain two problems and what the designer should have done.
    Check your answer
    Answer: It reveals an identifiable person’s health information, and it uses real data where illustrative data would do. The designer should use synthetic or illustrative data, or at least remove identifiers.
    De-identification removes identifiers such as names, most dates and small geographic units; even then, combining datasets can sometimes re-identify people.
  4. For the de-identified scans dilemma, write Position B’s strongest case (patients should be asked and able to say no) in 2-3 sentences, then restate it in one sentence a holder would accept.
    Check your answer
    Answer: Model case: Patients trusted the hospital with their scans for their own care, not for a company’s product, and combining datasets can sometimes re-identify people, especially when a condition is rare. Asking for consent respects that trust and lets people protect themselves. Restatement: ‘Because de-identification is not a perfect guarantee and the use is new, patients deserve to be told and to choose.’
    A strong answer uses a fact from the lesson (re-identification risk) and restates fairly before any reply.

Challenge

Stretch problems. Expect to think before you write.

  1. A chart of an invented 30-member club’s health survey shows one column: ‘age 60+, female, rare condition: 1’. Even with no names, why is this a risk, and how would you redesign it?
    Check your answer
    Answer: A group of one can single out an identifiable member, especially with a rare condition. Redesign by aggregating categories (combine ages or genders, or group rare conditions as ‘other’) and suppressing counts below a set minimum.
    Small groups and rare conditions are exactly where re-identification happens.

Review

Keep earlier skills sharp.

  1. Review 2.3: Using the bias checkpoint ‘who is in the training data’, explain why a company receiving scans from only one hospital should ask that question.
    Check your answer
    Answer: Patients at one hospital may not represent everyone the tool will be used on, so the model may work less well for groups who are missing.
    Who is in the data, and who is missing, is the first of the five checkpoints.

Missouri Learning Standards — Media Arts addressed: MA:Re9A.II, MA:Re8A.II, MA:Cn11A.II, SL.11-12.1, SL.11-12.3, RST.11-12.8

Instructional focus: Critique of data media in ethical and legal context

Reading

Health Data and Privacy: HIPAA in the Regulations’ Words, Missouri Genetic-Testing Law, Two Dashboards to Critique and the Scan Dilemma as a Written Debate

45 CFR 160.103 and 164.514 and 164.524 (eCFR, retrieved Sept. 26, 2026); HHS OCR, Health App Use Scenarios and FTC Health Breach Notification Rule guidance (paraphrased); NHGRI, Genetic Discrimination (updated Jan. 6, 2022); Revised Statutes of Missouri section 375.1303 (revisor.mo.gov, paraphrased, retrieved Nov. 10, 2026); SLS original dashboards, dilemma and positions · Public domain (U.S. federal regulations and agency pages), with the Missouri statute paraphrased and SLS original dashboards, dilemma and worked critique

Why this reading. The lab asks you to critique “two health dashboards or data graphics,” write a structured debate on a hospital-scan dilemma, and use HIPAA and other laws, and it supplies neither the graphics, the dilemma’s details nor the legal wording. This reading gives the federal text you need (quoted from the regulations themselves), a Missouri genetic-testing law in plain words, two dashboards described closely enough to critique, a privacy criterion with arithmetic, and the dilemma with two example positions and a debate frame. The two dashboards, the dilemma and the positions are SLS originals; every patient, clinic, count and name in them is invented. This is a description of the rules, not legal advice.

1. What HIPAA covers, from the regulations

Who is covered. Under 45 CFR 160.103, a covered entity is (1) a health plan, (2) a health care clearinghouse, or (3) “a health care provider who transmits any health information in electronic form in connection with a transaction covered by this subchapter.” A business associate is generally a person who, on behalf of a covered entity, creates, receives, maintains or transmits protected health information for a regulated function or activity, such as claims processing, data analysis, quality assurance or billing. What is protected. Protected health information is individually identifiable health information, which the regulation defines as information, including demographic information, created or received by a health care provider, health plan, employer or clearinghouse that relates to a person’s past, present or future health or condition, the provision of care or payment for care, and that identifies the person or could reasonably be used to identify them. Rights. Under 45 CFR 164.524, an individual has a right to inspect and obtain a copy of their protected health information in a designated record set, with a few exceptions (for example psychotherapy notes). (eCFR, retrieved 26 September 2026; U.S. government works, public domain.)

What HIPAA does not cover. If an organization is neither a covered entity nor a business associate, the HIPAA rules do not apply to it. HHS’s Office for Civil Rights has said that when a consumer downloads a health app that is neither, the information the app holds is no longer subject to HIPAA. The Federal Trade Commission says many companies that collect health information, such as a fitness tracker, a diet app or a connected blood pressure cuff, are not covered by HIPAA, and its Health Breach Notification Rule requires certain such organizations to notify customers, the FTC and sometimes the media after a breach of unsecured health information. (HHS OCR, Health App Use Scenarios; FTC business guidance; retrieved 26 September 2026 through their search listings; I paraphrase them.)

2. De-identification: the 18 identifiers

45 CFR 164.514(b) gives two ways to treat information as de-identified. Expert determination: a person with appropriate statistical knowledge determines that the risk is very small that the information, alone or combined with other reasonably available information, could identify a person, and documents the analysis. Safe harbor: remove the following identifiers of the person and of relatives, employers and household members, and have no actual knowledge that what remains could identify someone:

  1. Names
  2. Geographic units smaller than a state (street address, city, county, precinct, zip code), with a narrow exception for the first three digits of a zip code
  3. All elements of dates (except year) directly related to a person, such as birth, admission, discharge or death dates, and all ages over 89
  4. Telephone numbers
  5. Fax numbers
  6. Email addresses
  7. Social security numbers
  8. Medical record numbers
  9. Health plan beneficiary numbers
  10. Account numbers
  11. Certificate and license numbers
  12. Vehicle identifiers and serial numbers, including license plates
  13. Device identifiers and serial numbers
  14. Web addresses (URLs)
  15. Internet Protocol (IP) addresses
  16. Biometric identifiers, including finger and voice prints
  17. Full-face photographs and comparable images
  18. Any other unique identifying number, characteristic or code

Note the words “alone or in combination.” Removing names does not make a small group anonymous.

3. Genetic information and Missouri

The National Human Genome Research Institute explains that the Genetic Information Nondiscrimination Act of 2008 (GINA) protects Americans from discrimination based on genetic information in health insurance (Title I) and employment (Title II). GINA’s health insurance protections do not cover long-term care, life or disability insurance, though some states add protections (NHGRI page last updated January 6, 2022; retrieved 26 September 2026). Missouri is one of those states. Section 375.1303 of the Revised Statutes of Missouri says that an insurer, when deciding eligibility, premiums, limits or renewal for a health plan, may not require or request that a person or a blood relative provide genetic information or take a genetic test, and may not consider genetic information or test results without the person’s approval. The section defines “health plan” so that it does not include life insurance, disability income or long-term care coverage, and it has applied to coverage since January 1, 1999. Section 375.1300 defines genetic information as the results of a genetic test, and says it does not include family history or the results of routine physical measurements or blood, urine or chemical tests (Missouri Revisor of Statutes, sections 375.1300 and 375.1303, retrieved 10 November 2026; paraphrased). Genetic data is unusual because a result for one person also says something about biological relatives who never consented.

4. Two dashboards to critique (SLS original, invented)

Dashboard 1, “Riverbend Health Center: Flu Season at a Glance.” One landscape screen on a white ground. Left two-thirds: a map of one city divided by ZIP code, each ZIP shaded from pale yellow to dark red by number of flu visits; hovering a ZIP reveals a table by age group and sex. The ZIP 64xx7, in the north, is the darkest and reads “2 visits, ages 30–39, female.” Right third: a bar chart of visits by week with the number above each bar; below it a scrolling list titled “Recent visits” showing visit date, ZIP, sex, age and reason, most recent first. Title in 32-point navy, all else 10-point gray on white. Red is used both for “most cases” on the map and for the word “alert” on the list.

Dashboard 2, “My Year in Steps.” A tall phone-style screen. A large circular ring shows “Daily steps goal reached 251 days.” Below it, a line of resting heart rate by week with the highest week circled and labeled “Sick week? Data shows illness.” A card at the bottom compares “You vs. people in Springfield, ages 25–29” with four bars; a footer says “Shared with 12 other companies to improve your experience.” The screenshot shows the account name “Taylor R.” and email in the top corner.

The privacy criterion, with arithmetic

Ask of every graphic: could a viewer learn something about an identifiable person? Use these tests. (1) Smallest group. Find the smallest number of people any label describes; decide a minimum and state it (SLS suggestion for course work: fewer than 10 people is suppressed or merged). Dashboard 1 has a group of 2. (2) Combination. A label combining ZIP, sex and age band can name one person: if a ZIP has 4,000 residents, 2 sexes and 6 age bands make 12 groups of about 333 people on average, but a rare combination (here, women aged 30–39 in a small ZIP with only 2 flu visits) can be 1 or 2. (3) Identifiers on the screen. List every item that appears on the 18-identifier list, including dates and ZIP codes; Dashboard 1’s “Recent visits” list shows exact dates and ZIP codes. (4) Label claims. Does a caption assert a health condition the data cannot support? Dashboard 2 says “Data shows illness,” from a heart-rate rise. (5) Who else sees it. Dashboard 2’s footer says the data go to 12 companies; is it covered by HIPAA? Whether it is depends on whether the app works on behalf of a covered entity (see section 1).

Model critique (SLS original). On privacy, Dashboard 1 rates weak: the north ZIP shows 2 visits by a woman aged 30–39, a group of 2, and the recent-visits list carries exact dates and ZIP codes, both on the 18-identifier list. A fix: aggregate ZIPs to a region, suppress groups under 10, replace the visit list with weekly totals. Dashboard 2 rates weak: the screenshot carries a name and email, and a note asserting illness. A fix: crop the screenshot, replace the caption with “highest resting rate this year,” and state who receives the data.

5. The scan dilemma, in full (SLS original)

A hospital has 40,000 chest X-ray images with the patient identifiers removed, taken from patients over eight years, and wants to share them with a company that will train an AI tool to help radiologists read images. The company is not a covered entity. The hospital says the images are de-identified under the safe harbor list, so patient permission is not required; some patients say no one told them their images might be used this way. The company says the tool could help many patients; a patient advocate says images can still identify a person when combined with other data, especially for a rare condition.

Two example positions (invented; you will argue both): Position A: “Patients should be told in plain language at intake that de-identified images may be used for research, and they should be able to opt out, because trust depends on it.” Position B: “An opt-out could bias the data by leaving out the patients who most need the tool, so the hospital should use de-identification and oversight instead of asking each patient.” A fair restatement of B: “You think opt-outs could make the training data unrepresentative, and you would rely on de-identification plus review; is that right?” Then give your own position, with one reason that uses a fact from this reading (for example, that de-identification protects only if no one can combine what remains with other data).

6. The structured written debate (you argue both sides)

  1. Opening for A (about 80 words): A’s strongest reason, with one fact from this reading.
  2. Opening for B (about 80 words): B’s strongest reason, with one fact from this reading.
  3. A answers B: first restate B in a form a B supporter would accept (“You think…; is that right?”), then reply.
  4. B answers A: the same, the other way round.
  5. Your position (about 100 words): what patients should be told, whether they should be able to say no, and one reason.

If a family member is willing, they can read one side’s opening aloud and you answer it out loud before writing it down. Either way, judge your restatements by one test: would someone who holds that view sign it?

Sources: 45 CFR 164.514, Other requirements relating to uses and disclosures of PHI (eCFR) (ecfr.gov); 45 CFR 160.103, Definitions (eCFR) (ecfr.gov); 45 CFR 164.524, Access of individuals to PHI (eCFR) (ecfr.gov); HHS OCR, Health App Use Scenarios & HIPAA (hhs.gov); FTC, Complying with the Health Breach Notification Rule (ftc.gov); NHGRI, Genetic Discrimination (genome.gov); Revised Statutes of Missouri, section 375.1303, Genetic testing by insurer, limitations (Missouri Revisor of Statutes) (revisor.mo.gov); Revised Statutes of Missouri, section 375.1300, Definitions (Missouri Revisor of Statutes) (revisor.mo.gov)

Up NextComplete & Continue →

Leave a Reply Cancel reply

Pulse to Pixels: Coding a Heart-Rate Visualization
Prev
Imaging the Invisible: Designing a Patient Brochure
Next

© 2026 The VR School | Career content by Wiley | Powered by Sejal Learning Systems.

Join free · Accreditation · Privacy Policy · Terms of Service

Modal title

Main Content