
build and check an illustrative project budget, identify a plausible funding route and FDA pathway, analyze the ethics of my capstone, evaluate two invented projects as a review panel member, and answer the published panel questions on a recording.
Instruction
Real biomedical innovations must survive more than a design critique. They need money, they may need regulatory authorization, and they must be ethical. This course asks you to analyze financial and corporate considerations and ethical implications. Today you prepare that analysis, review two invented projects as a panel member would, and then face the panel’s questions about your own project on a recording.
Budgeting. A budget lists every cost with its unit price, quantity and total, then adds a contingency for the unexpected, often about 10%. Here is an illustrative six-month pilot budget for a student-designed app (every rate is invented for teaching). Developer time: 120 hours × $35 per hour = $4,200. Cloud computing: $150 per month × 6 months = $900. Usability-test thank-you gift cards: 12 participants × $25 = $300. Prototype materials: $180. Subtotal: $4,200 + $900 + $300 + $180 = $5,580. Contingency at 10%: $558. Total: $6,138. Always show the calculation, label rates as illustrative unless you have a source, and check the arithmetic twice.
Funding routes. Early biomedical projects are funded in several ways: research grants, including the NIH’s Small Business Innovation Research (SBIR) and Small Business Technology Transfer (STTR) programs for small companies; foundation grants; university and hospital innovation programs; investors, who take a share of the company in exchange for money and expect a return; and crowdfunding. Each brings obligations and influences what gets built: investors want growth, grant funders want public benefit and reporting.
Regulatory pathways. In the U.S., software that is intended to diagnose or treat disease can be a medical device regulated by the FDA; some low-risk wellness software is not. Devices reach the market mainly through three pathways. 510(k) clearance shows a device is substantially equivalent to a legally marketed device (a predicate). The De Novo pathway is for novel low- to moderate-risk devices with no predicate; IDx-DR used it in 2018. Premarket approval (PMA) is the most demanding pathway, for high-risk devices. Which one would fit your capstone if it were real? Say why, and say you are not certain; regulatory decisions depend on details.
Ethics. Research that tests a device on people requires protections. Under federal rules, most such research is reviewed by an institutional review board (IRB), which checks that risks are minimized and that participants give informed consent, understanding what will happen and that they can withdraw. Your ethics analysis should also cover the issues from this year: whose data trains the AI and whether they consented, how the system could be biased and for whom, privacy and HIPAA, what happens when it is wrong, and who is accountable.
The panel. A review panel has four roles: finance (is the budget complete and correct?), regulation (is the pathway reasoning sound?), ethics (are the key risks identified and addressed?) and design (does the concept meet the need?). First, sit on the panel: take each role in turn and write a defensible review of the reading’s two invented projects, each with a rating, evidence and one specific recommendation. Then sit in the presenter’s chair: record your answers, about 60 seconds each, to the eight published panel questions in the reading about your own capstone (a family member may read the questions aloud). After a day away, play the recording back, score it with the review form, restate the most important recommendation you would give yourself, and respond to it.
📖 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.
An illustrative budget has costs of $2,400, $600, $250 and $150. With a 10% contingency, what is the total?
A novel, moderate-risk AI screening device has no similar device already on the market. Which FDA pathway is most likely to fit?
Informed consent means participants understand what the research involves and know they can withdraw.
Write your panel review of one of the reading’s two invented projects in one role: a rating, evidence from the project description and one specific recommendation. Then, from your recorded answers to the published panel questions, restate the most important recommendation for your own capstone and your response.
How confident are you that you can build a correct illustrative budget, reason about funding, regulation and ethics, and review projects, including my own, rigorously?
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.
- Match each FDA pathway to its use: 510(k); De Novo; PMA. Uses: novel low- to moderate-risk device with no predicate; high-risk device, most demanding; device substantially equivalent to a legally marketed predicate.
Check your answer
Answer: 510(k): substantially equivalent to a predicate. De Novo: novel low- to moderate-risk device with no predicate (IDx-DR used it in 2018). PMA: high-risk devices, most demanding.Some low-risk wellness software is not regulated as a device; regulatory decisions depend on details. - Define contingency, IRB and informed consent.
Check your answer
Answer: Contingency: an amount added to a budget, often about 10%, for unexpected costs. IRB: an institutional review board, a committee that reviews research with human participants to protect their rights and welfare. Informed consent: a participant’s voluntary agreement after understanding what the research involves and that they can withdraw.IRBs check that risks are minimized and that participants give informed consent. - Name the four roles on the review panel and the question each asks.
Check your answer
Answer: Finance: is the budget complete and correct? Regulation: is the pathway reasoning sound? Ethics: are the key risks identified and addressed? Design: does the concept meet the need?A defensible review in each role gives a rating, evidence and one specific recommendation.
Application
Use the skill in context. Show your reasoning.
- Check the lesson’s illustrative budget: developer time 120 h x $35; cloud $150 x 6 months; 12 gift cards x $25; materials $180; 10% contingency. Compute each line, the subtotal and the total.
Check your answer
Answer: $4,200; $900; $300; $180; subtotal $5,580; contingency $558; total $6,138.120 x 35 = 4,200; 150 x 6 = 900; 12 x 25 = 300; 4,200 + 900 + 300 + 180 = 5,580; 10% of 5,580 = 558; 5,580 + 558 = 6,138. All rates are invented for teaching. - An invented project’s budget lists: developer time 80 h x $40 = $3,000; cloud $100 x 6 = $600; materials $200; contingency 10% = $380; total $4,180. As the finance reviewer, find the error and give the corrected total.
Check your answer
Answer: Developer time is wrong: 80 x $40 = $3,200, not $3,000. Corrected subtotal $4,000; contingency $400; total $4,400.3,200 + 600 + 200 = 4,000; 10% = 400; 4,000 + 400 = 4,400. The listed contingency of $380 matched the wrong subtotal of $3,800. - An invented capstone is an app that flags possibly worrying skin-spot photos for a clinician to review. If it were real and intended to help diagnose, which FDA pathway might fit and why? Say what you are not certain of.
Check your answer
Answer: Model: If no similar device were already legally marketed, De Novo might fit for a novel low- to moderate-risk device; if a similar device exists, 510(k). I am not certain: the risk level and the existence of a predicate depend on details, and higher risk could require PMA.Software intended to diagnose can be a medical device; the lesson asks you to give a reason and state your uncertainty. - Write a defensible ethics review of an invented project, ‘ClinicCheck’, which trained its photo classifier on images from one city hospital and did not ask the people in the photos for consent. Give a rating, evidence and one recommendation.
Check your answer
Answer: Model: Rating: weak. Evidence: the training images came from one hospital, so the classifier may work less well for groups missing from that population, and the people in the photos were not asked to consent. Recommendation: use only data collected with consent, test on images from several sites, and report performance for each group before any pilot.The ethics analysis covers whose data trains the AI and whether they consented, bias, privacy, what happens when it is wrong and who is accountable. - Compare two funding routes from the lesson, investors and research grants such as SBIR: what each expects, and how it might shape what gets built.
Check your answer
Answer: Investors take a share of the company and expect a return, so they push for growth. Grant funders (for example NIH SBIR and STTR for small companies) want public benefit and reporting, so they favor work with clear public value and documented results.Each funding route brings obligations and influences what gets built.
Review
Keep earlier skills sharp.
- Review 3.3: A patent and an FDA authorization are different things. In one sentence each, say what each one gives an inventor.
Check your answer
Answer: A patent gives a limited period of exclusive rights (generally 20 years from filing for a utility patent) in exchange for public disclosure. An FDA pathway such as 510(k) or De Novo allows a medical device to reach the U.S. market.A patented device may still need FDA authorization, and an authorized device need not be patented.
Review Panel: Two Projects, a Budget Check, the Rules, a Review Form and the Published Panel Questions
FDA pages on 510(k), De Novo and PMA and the FDA Digital Health Advisory Committee executive summary (retrieved 2026-09-26); HHS OCR; 45 CFR 46.116 (eCFR 2026-09-26); NIST AI 100-1 (Jan. 2023); Projects P and Q, budgets, review form and panel questions are SLS original and illustrative · Public domain (FDA, HHS, NIST, 45 CFR 46); SLS original (invented projects, budgets, review form)
Why this reading. A review panel needs projects to review, a way to check a budget, the rules a reviewer applies (FDA pathway, consent, risk), a review form and the questions a panel asks. This reading supplies all five, so you can work with nothing but this page: two invented projects to review in writing, the arithmetic worked, quoted rules from FDA, HHS, NIST and the federal human-subjects regulation, a model review, and the published panel questions you answer about your own capstone on a recording.
1. Checking a budget
| Line | Quantity x rate | Total |
|---|---|---|
| Developer time | 120 hours x $35 | $4,200 |
| Cloud computing | 6 months x $150 | $900 |
| Thank-you gift cards | 12 x $25 | $300 |
| Prototype materials | $180 | |
| Subtotal | $4,200 + $900 + $300 + $180 | $5,580 |
| Contingency at 10% | 0.10 x $5,580 | $558 |
| Total | $5,580 + $558 | $6,138 |
Illustrative rates, invented for teaching. Second check, the quiz-sized budget: $2,400 + $600 + $250 + $150 = $3,400; 10% is $340; total $3,740.
2. Regulatory pathways: what FDA says
- 510(k). A premarket submission that demonstrates the device is as safe and effective, that is, “substantially equivalent,” to a legally marketed device (a predicate). (FDA, Premarket Notification 510(k))
- De Novo. A pathway to classify novel devices for which general controls, or general and special controls, provide reasonable assurance of safety and effectiveness, but for which there is no legally marketed predicate device. (FDA, De Novo Classification Request). FDA reviewed IDx-DR under De Novo; the device is DEN180001, final decision April 11, 2018.
- Premarket approval (PMA). “The FDA process of scientific and regulatory review to evaluate the safety and effectiveness of Class III medical devices,” which are devices that support or sustain human life, are of substantial importance in preventing impairment of human health, or present a potential unreasonable risk of illness or injury. (FDA, Premarket Approval)
- Wellness software outside FDA review. FDA’s advisory-committee executive summary notes that the majority of commercially available mental-health apps are consumer wellness apps that “are not reviewed or authorized by FDA,” and that FDA applies oversight to software functions that are medical devices and whose functionality could pose a risk to a patient’s safety if the device did not function as intended.
| Practice case (invented) | Likely pathway, and the one thing that would change it |
|---|---|
| A phone app that counts steps and encourages walking | Consumer wellness; not reviewed. Changes if it claims to diagnose or treat a disease. |
| Software that flags lung nodules on CT scans, similar to tools already on the market | 510(k) if a legally marketed predicate exists. Changes if its intended use or technology raises different questions. |
| Software that gives a screening decision for an eye disease with no similar device on the market | De Novo, as for IDx-DR in 2018. Changes if the risk is judged high. |
| An implanted device that sustains life | PMA. Class III risk. |
Say which pathway your capstone would follow if it were real, say why, and say you are not certain: the decision depends on the device’s intended use.
3. Ethics: consent and the rules for research on people
Under the federal Common Rule (45 CFR 46.116(b)), an investigator seeking informed consent must provide each subject with, among other things: a statement that the study involves research, its purposes, expected duration and procedures; a description of reasonably foreseeable risks or discomforts; a description of expected benefits; alternatives; the extent to which confidentiality will be maintained; whom to contact with questions; and “a statement that participation is voluntary, refusal to participate will involve no penalty,” and that the subject may stop at any time. An institutional review board (IRB) reviews such research. A home think-aloud test with fictional scenarios, run with adults you know who agreed to take part, differs from research on patients; still, ask who your testers are, what you tell them and what you keep.
Privacy. HHS states that the HIPAA Rules apply to covered entities and their business associates; an app that is neither is outside HIPAA. A capstone that stores health readings should say what it stores and where.
4. A risk frame: NIST AI RMF 1.0 (NIST AI 100-1, January 2023, public domain)
The framework’s core has four functions: GOVERN (cultivates a culture of risk management and outlines processes to anticipate, identify and manage the risks a system can pose, including to users and others); MAP (“establishes the context to frame risks related to an AI system”); MEASURE (“employs quantitative, qualitative, or mixed-method tools, techniques, and methodologies to analyze, assess, benchmark, and monitor AI risk”); and MANAGE (“entails allocating risk resources to mapped and measured risks”). Use them as four questions for a project: who is responsible, what is the context and who could be harmed, how was it tested, and what happens when it fails.
5. Two projects to review (SLS original, invented)
Project P: CoughCheck. A phone app that listens to a 10-second cough recording and labels it “typical” or “see a doctor.” Team claim: “It needs no FDA review because it is just an app.” Data: 40 volunteer recordings, no consent form, stored on a shared drive. Test: 30 recordings; the app labeled 27 of 30 the same as the team’s own listening (90%). Budget as submitted: developer time 80 hours x $30 = $2,400; microphone kit $120; cloud 5 months x $60 = $300; gift cards 10 x $20 = $200; subtotal $3,120; contingency 10% $312; total $3,432.
Project Q: Dose-Check Tray. The tray from Lesson 3.4. Team claim: “We expect 510(k) clearance soon.” Data: no patient data; four users tested a paper prototype with fictional scenarios, each told they could stop. Test: 2 rounds; 5 of 7 compartments read correctly in dim light, 7 of 7 with a lamp. Budget as submitted: 4 prototypes at $85 = $340; 4 camera modules at $22 = $88; 12 thank-you gifts at $15 = $180; printing $60; subtotal $668; contingency 10% $66.80; total $734.80.
| Reviewer check | Project P | Project Q |
|---|---|---|
| Arithmetic | Wrong: the lines sum to $3,020; 10% is $302; the correct total is $3,322 (the submitted $3,432 is $110 too high) | Correct: 340 + 88 + 180 + 60 = 668; 10% = 66.80; total 734.80 |
| Regulatory claim | “Just an app” is not a test: the deciding fact is the intended use. A label that says “see a doctor” on a cough sound is a health claim | “510(k) soon” assumes a predicate exists; nothing in the project shows one |
| Consent and privacy | No consent form; recordings on a shared drive; ask about voluntary participation and confidentiality | Fictional scenarios, voluntary, stop at any time: consistent with the consent elements |
| Test | Agreement with the team’s own listening is not a test against a clinical standard; 30 recordings is small | Two rounds with a revision between them; small sample stated? |
6. Review form and a model review
| Part | What to write |
|---|---|
| Rating (1-4) | Overall, against: honest budget, regulatory reasoning, ethics, test evidence |
| Evidence from the project | Quote or paraphrase two things the project description says |
| One specific recommendation | A change the team could make this week |
Model review of Project P. Rating 2. Evidence: the four budget lines add to $3,020 ($2,400 + $120 + $300 + $200), not the submitted $3,120, so the total should be $3,322, not $3,432; and the claim “needs no FDA review because it is just an app” is stated as fact. Recommendation: state the intended use in one sentence, drop the “see a doctor” label or say plainly that FDA review might apply, and write a consent form covering what the recordings will be used for and who can hear them.
7. The published panel questions (answer them about your own capstone, on a recording)
A real review panel would question you aloud. Here the panel’s questions are published so you can prepare. Record yourself answering each in about 60 seconds (audio, or a screen recording of your slides; your face need not appear). A family member may read the questions aloud to you. Keep the recording private unless a parent agrees to share it. After a day away, listen back and score each answer 1 to 4 with the review form above.
| Role | Question |
|---|---|
| Finance | 1. Walk me through your budget. Do the lines add to the subtotal, and what is the contingency? |
| Finance | 2. Which rates have a source, and which are illustrative? Which funding route fits, and what would that funder expect in return? |
| Regulation | 3. State the intended use in one sentence. Which FDA pathway, if any, would likely apply, and why? |
| Regulation | 4. What single fact about your design would change that pathway? |
| Ethics | 5. Whose data would train or test the AI, and how would they give informed consent? |
| Ethics | 6. Who could be harmed by a wrong output, and what in the design limits that harm? |
| Design | 7. What user need does it meet, and what did your testing show? |
| Design | 8. What is the panel’s most important recommendation likely to be, and how would you respond? |
Sources: FDA, Premarket Notification 510(k) (fda.gov); FDA, De Novo Classification Request (fda.gov); FDA, Premarket Approval (PMA) (fda.gov); 45 CFR 46.116, General requirements for informed consent (ecfr.gov); NIST, Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1 (doi.org)