
develop, test and exhibit an original biomedical AI innovation with a prototype, a recorded TED-style talk and a professional portfolio, and respond to visitor feedback and my own review.
Instruction
This is the culmination of the year. You will bring an original biomedical AI innovation from concept to an exhibition you curate at home, with a catalog that can go online only with a parent’s permission. Everything you have learned is in play: how machine learning works and fails, the body systems and their signals, the history and ethics of AI in medicine, testing and honest accuracy, engineering design and patents, and above all the craft of making media that communicates.
What you will exhibit. A prototype, tangible or digital, at the fidelity your testing supports, with a demonstration visitors can try or watch. A recorded TED-style talk of five to seven minutes explaining the healthcare challenge, your design process, how the innovation works (including the AI component and where humans stay in control), and your technical and aesthetic choices. And a professional portfolio with research findings, design plans, a hypothetical patent section (drawings and claims), an illustrative budget with every figure calculated, a regulatory and ethics analysis, and a written reflection.
Testing requirement. You must test your prototype at least twice, with a change between tests that responds to what you found. Report results honestly, with numbers where you have them and without claims your tests cannot support. Your innovation is a concept and prototype, not a medical device; say so on your exhibit.
Designing the exhibit. Set it up at home on a table or wall for family members and other adults you know, and make a PDF or slide catalog of it, one page each for the prototype, process, results and limits, with your recorded talk. Share the catalog online only if a parent agrees, without personal details. Apply what you learned from the Eames Office: give visitors something to do, make one object the center, show your process (a strip of photographs from sketch to prototype works well), and be honest about limits. Carry one visual identity across talk slides, exhibit signage and portfolio.
Feedback at the exhibition. Visitors will leave structured feedback on cards: one strength, one question, one suggestion. You will also score your recorded rehearsal against the rubric and the reading’s self-review sheet after a day away. In your reflection, you will analyze this feedback, identify its patterns and say what you would change next.
The reflection. About 900 words: how your idea changed from proposal to exhibit and why; the most important thing each test taught you; one ethical tension in your design and how you handled it; how the feedback confirmed or challenged your view of the work; and what the year has taught you about designing AI for people.
The four stages. Stage 1: the capstone plan, including the proposal from Lesson 4.1, the first test results from Lesson 4.2 and the budget and ethics analysis from Lesson 4.3. Stage 2: a full recorded rehearsal of the talk and a draft portfolio, which you critique after a day away with the self-review sheet (a family member or other adult may score it too); submit the notes and your revision plan. Stage 3: the revised final prototype, talk, portfolio and second test. Stage 4: the home exhibition and catalog, with your recorded talk and recorded answers to the three published visitor questions, followed by the reflection.
📖 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.
Formative check
Work through these before moving on. They are not graded — they tell you whether the standard below has landed yet.
Your second test showed users misread the confidence score. What belongs in your reflection?
The capstone exhibit should state that the innovation is a concept and prototype, not an authorized medical device.
How confident are you that you can develop, test, exhibit and reflect on an original biomedical AI innovation?
Unit 4 checkpoint
Do this before you start the task. It covers the whole unit and tells you what to review first.
Which exhibition taught through hands-on devices such as funhouse mirrors and a cone in which balls speed up near the center?
Exhibitions about the future are persuasive media that may carry a sponsor’s interests.
A question form that turns a problem into an opening for ideas begins ‘How we…’.
In a Wizard of Oz test, who produces the AI’s output?
A self-test after a day away may count as one user if no tester is available, as long as it is labeled in the log.
An illustrative budget has costs of $3,000, $500 and $500. With a 10% contingency, what is the total?
Which FDA pathway shows that a device is substantially equivalent to a legally marketed device?
A committee that reviews research with human participants is an institutional review .
In a STAR answer, which part describes what you actually did?
A portfolio should put the strongest piece first and be ordered to show growth.
Career information such as typical education and outlook comes from the Bureau of Labor Statistics Occupational Outlook .
Submit: capstone prototype and demonstration + recorded 5-7 minute talk + home exhibition catalog + portfolio (patent section, budget, ethics) + 900-word reflection
Task: Develop an AI-driven biomedical innovation that addresses a documented healthcare challenge. Test the prototype at least twice with a revision between tests. Exhibit it with a demonstration at a curated home exhibition with a PDF or slide catalog (online only with a parent’s permission), record a five- to seven-minute TED-style talk and submit a professional portfolio with research, design plans, a hypothetical patent section, an illustrative budget with calculations, a regulatory and ethics analysis, and a 900-word reflection that responds to visitor feedback.
Submission workflow
- Stage 1 — Capstone plan. Proposal, first test results, illustrative budget and regulatory and ethics analysis.
- Stage 2 — Self-critiqued draft. Full recorded talk rehearsal and draft portfolio, scored after a day away with the self-review sheet (a family member or other adult may score it too); submit the notes and your revision plan.
- Stage 3 — Revised final. Revised prototype after a second test, final talk, final portfolio.
- Stage 4 — Exhibition and reflection. Hold the home exhibition, publish the catalog (online only with a parent’s permission) with your recorded talk and recorded answers to the published visitor questions, collect feedback cards, then submit the 900-word reflection.
How it is scored
| Criterion | What proficient looks like |
|---|---|
| Transdisciplinary synthesis (MA:Pr4A.III) | Science, AI design, engineering, ethics and media art are synthesized into one compelling capstone that retains artistic fidelity across prototype, talk and portfolio. |
| Production mastery (MA:Cr3A.III) | Prototype, exhibit, slides and portfolio show purposeful, polished and original craft. |
| Exhibition design and delivery (MA:Pr6A.III, SL.11-12.4) | Exhibit, catalog and recorded talk are curated for the audience with intentional impact; delivery is clear and the published visitor questions are answered accurately. |
| Testing, evidence and honesty (WHST.11-12.2) | Two tests with a revision between them; results and limits reported accurately; budget arithmetic correct. |
| Ethics and context (MA:Cn11A.III) | Data consent, bias, privacy, accountability and regulation are analyzed and addressed in the design. |
| Reflection and response to feedback (MA:Re9A.III) | Rigorous self-evaluation that analyses patterns in feedback and states specific next steps. |
Capstone Exhibition: The Full Brief, an Exemplar of Two Tests, a Home Exhibition and a Worked Feedback Analysis
SLS original assignment brief, criteria, exemplar tests, home-exhibition guide, feedback-card analysis and self-review sheet (RLP at-home version, 2026-11-10); all counts are invented and illustrative · SLS original
Why this reading. The capstone task lists what to submit, but you need the brief itself: the assignment, criteria at four levels, an exemplar showing two tests with a revision between them, a worked analysis of exhibition feedback cards (so you know what “patterns in feedback” looks like), how to stage a home exhibition and catalog, and the self-review sheet with the published panel questions. The exemplar is invented; its numbers show the calculations, not a result to copy.
The assignment
- Develop an AI-driven biomedical innovation that addresses a documented healthcare challenge (a federal source, Reading 4.1).
- Test the prototype at least twice, with a change between tests that responds to what you found.
- Exhibit it with a demonstration at a curated home exhibition (see below); record a five- to seven-minute talk; submit a portfolio with research, design plans, a hypothetical patent section (drawings and claims), an illustrative budget with every figure calculated, and a regulatory and ethics analysis (Reading 4.3).
- State on the exhibit that the innovation is a concept and prototype, not an authorized medical device.
- Collect feedback cards (one strength, one question, one suggestion) from your visitors and write a 900-word reflection that analyzes them.
Staging the exhibition at home
- Home showing. Set up the prototype, a demonstration visitors can try, a process strip of photographs and a sign stating that the innovation is a concept and prototype. Invite family members and other adults you know; each visitor fills in a feedback card. Several visitors over a day or two is enough.
- Catalog. Make a PDF or slide catalog of the exhibit: one page each for the prototype, the process, the test results and the limits, plus the link or file of your recorded talk.
- Online only with permission. Share the catalog or talk online only if a parent agrees, with no full name, school, address or face you have not chosen to show.
- Talk. Record it. Then record answers to the three published visitor questions: “What does the AI do, and who checks it?” “How many people tested it, and what changed?” “What can it not do yet?”
Criteria at four levels (SLS original)
| Criterion | 2 Developing | 3 Proficient | 4 Exemplary |
|---|---|---|---|
| Transdisciplinary synthesis | Parts sit side by side | Science, AI, engineering, ethics and design in one project | One visual identity across prototype, talk and portfolio; each part explains another |
| Production mastery | Some craft | Prototype, slides and portfolio polished | Original craft choices explained |
| Exhibition design and delivery | Reads from slides | Exhibit invites a visitor to do something; talk 5-7 minutes | Answers accurately; admits limits |
| Testing, evidence and honesty | One test | Two tests with a revision between; limits stated; budget correct | Reports what the second test did not fix |
| Ethics and context | Mentioned | Consent, privacy, bias, accountability and regulation each addressed | Design changed because of an ethical tension |
| Reflection and feedback | Lists comments | Groups comments into patterns | Says what each pattern changes next |
Exemplar: two tests with a revision (invented, SLS original)
Prototype: a paper reading coach (Reading 4.1, concept 1). Test 1, five users, fictional readings: 2 of 5 misread the on-screen “confidence 0.82” as “82% chance my blood pressure is high.” Revision: replaced the number with the words “Likely high. Recheck.” Test 2, five different users: 0 of 5 misread it. Reported honestly: “Misreadings fell from 2 of 5 (40%) to 0 of 5. With five users per round, this shows the change is promising, not proven; a user of a different age or background might still misread it.” In the reflection: describe the finding, the change, and what it taught you about designing for uncertainty.
Exemplar: talk timing
At 130 to 150 words per minute, a five-minute talk is 650 to 750 words and a seven-minute talk is 910 to 1,050 words (130 x 5 = 650; 150 x 5 = 750; 130 x 7 = 910; 150 x 7 = 1,050).
Worked analysis of feedback cards (invented data)
Twelve cards were collected at a home showing and from parent-approved viewers of the online catalog. Each has a strength, a question and a suggestion, coded into short labels.
| Card | Strength | Question | Suggestion |
|---|---|---|---|
| 1 | one clear result | how many people tested? | say the limit on the board |
| 2 | color and layout | does it work in dim light? | bigger type |
| 3 | one clear result | what if the phone is old? | say the limit on the board |
| 4 | the demo | is it a real medical device? | say the limit on the board |
| 5 | color and layout | who checks the AI? | show the human check |
| 6 | the demo | how many people tested? | say the limit on the board |
| 7 | one clear result | does it store my data? | show the human check |
| 8 | the demo | what if it is wrong? | show the human check |
| 9 | color and layout | how many people tested? | bigger type |
| 10 | one clear result | is it a real medical device? | say the limit on the board |
| 11 | the demo | what if it is wrong? | show the human check |
| 12 | color and layout | does it work in dim light? | bigger type |
Counts. Strengths: one clear result, 4; color and layout, 4; the demo, 4 (total 12). Questions: how many people tested?, 3; does it work in dim light?, 2; is it a real medical device?, 2; what if it is wrong?, 2; what if the phone is old?, 1; who checks the AI?, 1; does it store my data?, 1 (total 12). Suggestions: say the limit on the board, 5; show the human check, 4; bigger type, 3 (total 12).
Pattern and next step (model). The most common question was “how many people tested?” (3 of 12, 25%), and the most common suggestion was “say the limit on the board” (5 of 12, 42%). Both point to the same gap: the exhibit board hides the size of the test and the limits. Next step: put “5 users per round, one setting, concept only” in large type in the center panel. A pattern needs several cards; one unusual card is a note, not a pattern.
Reflection outline: about 900 words
| Section | Words |
|---|---|
| How my idea changed from proposal to exhibit, and why | 200 |
| What each test taught me | 200 |
| One ethical tension in my design and how I handled it | 150 |
| What the feedback confirmed or challenged | 150 |
| What this year taught me about designing AI for people | 200 |
200 + 200 + 150 + 150 + 200 = 900.
Self-review sheet for Stage 2 (score your recorded rehearsal and draft portfolio after a day away; a family member or other adult may score it too)
| Question | Score 1-4 | Evidence |
|---|---|---|
| Do two tests appear, with a change between them that responds to the first? | ||
| Is every number in the budget, and every count in the results, correct when I recompute it? | ||
| Does the exhibit say it is a concept and prototype, not a medical device? | ||
| Is the human check visible in the design? | ||
| Does the talk fit 5 to 7 minutes (650 to 1,050 words)? | ||
| Most useful revision: |
Sources: CDC, High Blood Pressure Facts and Stats (cdc.gov)