This page contains the working behind the Health Observer Systems comparative review of the Medical Snapshot system: feature definitions, search and source notes, and the full scoring table. The R source reproduces the plots.
This is very boring but I’m showing how I got to my conclusion. If someone is into statistics and would enjoy improving this design then yes please I’d love to hear from you.
My research question
This analysis is an exploration of how existing systems compare to my hypothetical Snapshot system. The matrix measures how close each system comes to the Snapshot’s architecture, asking the narrow question: which of the Snapshot’s features has each existing system implemented?
The 18 features fall into six dimensions, ordered like this:
What is gathered?"] --> B["B: Information Embargo
Who sees it?"] B --> C["C: Retroactive Release
When does it come back?"] C --> D["D: Directionality
Who holds power?"] D --> E["E: Participation Ethics
What's the deal?"] E --> F["F: Governance
Who enforces the rules?"]
A: Data Collection (4 features). Does the system test broadly across many conditions, test healthy people, test the same people repeatedly over time, and retain biological samples? These are the operating mechanics and reasonably objective.
B: Information Embargo (3 features). Does the system withhold results from participants, operate independently from their clinical care, and keep the specific tests conducted unknown to participants? There’s a bit more judgement needed here.
C: Retroactive Release (3 features). Does a formal pathway exist for releasing stored data when something happens later? Is it triggered by the treating doctor’s diagnosis? Does the released data primarily benefit the individual? This can be quite hard to pin down.
D: Directionality (3 features). Can the participant or their doctor initiate release (or both)? Is state access structurally excluded? Did the participant choose to join? This isn’t necessarily documented, because it can be embedded in law (there’s no need to write down that “The Law Says The Government Trumps All”, because that is the context of society) or there can be evidence that medical discretion has been exercised regardless of policy. That’s why I called it a direction, asking “to what did the designers aspire?”
E: Participation Ethics (2 features). What flows back to the participant? E14 records the benefit-sharing variant: reimbursement of expenses, wage-like payment for time, a data dividend, community benefit, or no compensation at all. The Snapshot pays for time, so E14 scores 1 for wage-like payment, ½ where expenses alone are reimbursed, and 0 for the rest. Does commercial early-access funding support the operation?
F: Governance (3 features). Is eventual public data release mandatory? Do trustees have a legal duty to participants, or to funders or governments? Is data stored across multiple legal jurisdictions?
A score of 1 means the system shares that feature with the Snapshot. A score of 0 means it doesn’t. A score of −1* means the system does the opposite of what the Snapshot intends: compulsory participation where the Snapshot requires voluntary, or state control where the Snapshot requires exclusion. Inversions count as 0 in the total but are marked separately because they carry information.
The Scoring
Systems are ordered by total score descending, then by year of establishment.
| System | A1 | A2 | A3 | A4 | B5 | B6 | B7 | C8 | C9 | C10 | D11 | D12 | D13 | E14 | E15 | F16 | F17 | F18 | Total |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Medical Snapshot (reference) | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1[wage] | 1 | 1 | 1 | 1 | 18 |
| DoD Serum Repository (USA, 1985) | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 0 | 0 | 0 | 0[?] | 0 | 0 | 0 | 0 | 8 |
| EPIC (Europe, 1992) | 1 | 1 | 1 | 1 | 1 | ½ | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 7.5 |
| UK Biobank (2006) | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | ½[reimb] | 0 | 1 | 0 | 0 | 7.5 |
| China “Physicals for All” (2013–) | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 0 | −1*[C] | −1*[C] | −1* | −1*[C] | 0[?] | 0 | 0 | −1* | 0 | 7* |
| China Kadoorie Biobank (2004) | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 7 |
| Taizhou (China, 2009) | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 7 |
| All of Us (USA, 2018) | 1 | 1 | 1 | 1 | −1* | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 1[wage] | 0 | 1 | 0 | 0 | 7 |
| ALSPAC (UK, 1991) | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | ½[reimb] | 0 | 1 | 0 | 0 | 6.5 |
| Framingham (USA, 1948) | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 6 |
| Whitehall I (UK, 1967) | 1 | 1 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 6 |
| Nurses’ Health Study (USA, 1976) | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 6 |
| BioBank Japan (2003) | 1 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 6 |
| CNHBM (China, 2017) | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 0 | 0 | 0 | 6 |
| Guthrie Card (global, 1963) | 0 | 1 | 0 | 1 | ½ | 1 | 0 | 1 | 0 | 1 | 0 | 0 | 0 | 0[none] | 0 | 0 | 0 | 0 | 5.5 |
| Generation Scotland (2006) | 1 | 1 | 0 | 1 | ½ | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 5.5 |
| deCODE (Iceland, 1998) | 1 | 1 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | −1* | 0[none] | 1 | 0 | 0 | 0 | 5 |
| Estonian Biobank (2001) | 1 | 1 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 1 | 0 | 0 | 5 |
| FinnGen (Finland, 2017) | 1 | 1 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 1 | 0 | 0 | 0 | 5 |
| Our Future Health (UK, 2022) | 1 | 1 | 0 | 1 | −1* | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0[?] | 0 | 0 | 0 | 0 | 4 |
| China CNGB / BGI GeneBank (2016) | 0 | 1 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | −1* | 0 | 0[?] | 0 | 0 | 0 | 0 | 4 |
| Majengo Cohort (Kenya, 1985) | 0 | 0 | 1 | 1 | ½ | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ½ | 0[comm] | 0 | 0 | 0 | 0 | 3 |
| NDNAD / CODIS (UK/USA, 1995/1998) | 0 | −1* | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | −1* | −1* | 0[none] | 0 | 0 | 0 | 0 | 3 |
| China MPS DNA database (2003) | 0 | −1* | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | −1* | −1* | 0[none] | 0 | 0 | −1* | 0 | 3 |
| Tuskegee (USA, 1932) | 0 | −1* | 1 | 0 | −1* | −1* | 1 | 0 | 0 | −1* | −1* | −1* | −1* | 0[comm] | 0 | 0 | −1* | 0 | 2 |
| NZ Unfortunate Experiment (1966) | 0 | −1* | 1 | 0 | −1* | −1* | 0 | 0 | 0 | −1* | −1* | −1* | −1* | 0[none] | 0 | 0 | −1* | 0 | 1 |
Notation: 1 = feature present. 0 = absent. ½ = partial. −1* = inverted (system does the opposite); counts as 0 in total. −1*[C] = inverted but evidence rests on advocacy-adjacent inference only. ASPI excluded at all tiers.
E14 benefit-sharing variants: [wage] wage-like payment for the participant’s time, which is what the Snapshot does. [reimb] expenses reimbursed and nothing further. [div] a data dividend. [comm] community or in-kind benefit. [none] nothing. [?] not yet verified, and scored 0 by default, so fifteen of these are placeholders. Individual in-kind inducements are recorded under [comm] for want of a closer variant, which is where Tuskegee’s free meals and burial insurance sit.
Also, I used “½” because it’s easier for a human to scan although it isn’t as computable.