Parth TiwariStart a project
MenuClose

Case 08 / Source-backed cancer education

OncoVerse

OncoVerse uses anatomy as an educational interface, but content completion and medical review state decide what the atlas is allowed to reveal.

Read the evidence record View repository
Role
Product engineer
Built at
Personal project
Contribution
Atlas concept, content model, sourcing, and review states
Status
In progress
The committed OncoVerse atlas artwork used by the source-backed cancer education interface
Committed atlas assetMain at f8bdbd3

Measured, with the denominator attached

Proof before polish.
1 complete of 5

OncoVerse currently has 1 complete atlas entry across a 5-entry directory, and the complete entry still needs medical review.

5 cancer directory entriesVerified 2026-08-28
What it is

01 / What it is

The anatomy is an interface. The review state is the guardrail.

OncoVerse is an educational atlas for patients, families, students, and educators.

It connects anatomy and plain-language explanations to source-backed content rather than offering diagnosis or treatment advice.

Built for

Patients, families, students, and educators looking for plain-language cancer education tied to visible anatomy and explicit source and review boundaries.

I designed the atlas concept, structured content directory, source metadata, review-state contract, Medullary Thyroid Carcinoma entry, and the acceptance path that keeps unfinished medical content from looking complete.

Contribution record

02 / Product proof

Read the source evidence before the argument.

Committed content inventory

Committed content inventoryThe atlas shape is ahead of the content
Directory entries
Five
Complete entries
One
Complete topic
Medullary Thyroid Carcinoma
Medical review
Still required
Source: Committed main at f8bdbd3
This is a readable rendering of the committed content directory. It is the primary product proof because the current uncommitted anatomy work is outside the audited source boundary.
01
The committed OncoVerse atlas illustration used as the anatomy-led navigation surface
Choose anatomy as the index

The atlas connects a body region to plain-language disease material instead of beginning with a dense article list.

02
Directory recordVisibility follows content state
Complete
May render with its sources
Stub
Remains visibly incomplete
Review pending
Cannot appear medically approved
Source: Committed content directory
Read structured content state

Each directory entry carries completeness, sourcing, and medical-review fields that the interface must obey.

03
Acceptance recordComplete is not reviewed
Topic
Medullary Thyroid Carcinoma
Content state
Complete
Review state
Needs medical review
Source: Committed MTC content entry
Accept content through a gate

A topic moves from draft to complete only after its structured sections and sources exist, then retains a separate medical-review state.

03 / Problem and responsibility

A beautiful health atlas can make incomplete content look medically complete.

Immersive health graphics can make thin content feel authoritative.

The product needs to show what is sourced, what is incomplete, and what still needs medical review without letting visual polish erase those boundaries.

Product
Anatomy-led navigation, educational reading sequence, boundary copy, and incomplete-state treatment.
Content system
Structured cancer entries, source metadata, completeness, review state, and acceptance rules.
Research
Medical source boundaries, visual-authority risk, atlas feasibility, and staged 3D exploration.

04 / Research that changed the build

Research only matters here when it changed a decision.

  1. 01

    Anatomy-as-interface studies

    Spatial context can make disease location easier to understand than a flat directory, but it also increases perceived authority.

    Anatomy became navigation while source and review state stayed visible in the reading layer.
  2. 02

    Content inventory audit

    The ambitious atlas shape contained one complete entry and four stubs.

    The site now publishes the exact scope instead of letting the interface imply a full cancer catalogue.
  3. 03

    Medical source-boundary review

    Educational explanations need explicit sourcing and must not drift into diagnosis or treatment recommendation.

    Each entry carries sources and the product refuses report interpretation, diagnosis, and treatment advice.
  4. 04

    Model-acceptance pipeline design

    Visual completion, content completion, and medical review are separate states.

    The content model keeps those states separate so a polished render cannot promote an unreviewed entry.

05 / Architecture

Treat content status and medical review as data that controls what the atlas may reveal.

Each directory entry carries structured content, source references, completion state, and review state.

The visual layer must read those fields directly so incomplete or unreviewed material cannot accidentally look finished.

Directory state controls what the atlas may expose
  1. 01DirectoryLists cancer topics, anatomy mapping, completion, and review state
  2. 02Content entryStores plain-language sections and attached medical sources
  3. 03AcceptanceChecks required structure without claiming medical approval
  4. 04Atlas surfaceReveals only what the content and review fields permit
  5. 05Reader boundaryEducation only, with no diagnosis or treatment recommendation

06 / Decisions and trade-offs

What I chose, what I rejected, and what the choice costs.

DecisionRejectedTrade-off
Let content and review state control visibility.

Render every atlas region and add real content later.

The atlas looks sparse, while incomplete medical material cannot masquerade as finished education.

Keep the current case study static and paper-first.

Use Three.js polish as proof that the educational product is complete.

The page is less immersive, while proof remains tied to committed content rather than a world prototype.

Separate complete from medically reviewed.

Treat a filled content schema as equivalent to expert approval.

Even the strongest entry retains a warning until qualified review exists.

07 / Evidence

One entry is complete out of five, and it still needs medical review.

1 complete of 5

OncoVerse currently has 1 complete atlas entry across a 5-entry directory, and the complete entry still needs medical review.

5 cancer directory entriesVerified 2026-08-28

The published 1-of-5 result is a repository content inventory. It describes scope, not medical quality, readership, completion velocity, or clinical validation.

08 / What failed

The failures are part of the architecture record.

  1. 01

    The atlas promised more than the repository contained

    What I saw
    A broad visual catalogue suggested many finished cancer topics.
    Why
    The product shape and visual research advanced before the source-backed content inventory.
    What changed
    The current scope is published as one complete entry across five, with four stubs left visibly incomplete.
    Still true
    Adding regions faster than sourced content would recreate the same mismatch.
  2. 02

    Complete could be mistaken for reviewed

    What I saw
    A structurally filled entry could look medically approved even while review was pending.
    Why
    Content completeness and expert review were initially easy to collapse into one status.
    What changed
    The model now keeps completion and medical review as separate fields and interface states.
    Still true
    The MTC entry still needs qualified medical review.
  3. 03

    Three-dimensional polish became a shortcut

    What I saw
    3D exploration risked becoming the visible proof before content, accessibility, and medical review were ready.
    Why
    The immersive surface was easier to showcase than the less glamorous content pipeline.
    What changed
    This case study uses committed static evidence; Three.js remains a later world decision.
    Still true
    Future 3D work still needs performance, accessibility, and source-state parity gates.
Read the content-scope correction

09 / Limits and future

What the product cannot do yet, beside what may come next.

Current limitations

  • Only one of five directory entries is complete, and that Medullary Thyroid Carcinoma entry still needs medical review.
  • OncoVerse is educational only and is not a diagnostic system, treatment recommender, or medical-report interpreter.
  • The current static proof does not establish that the full atlas interaction or 3D research is production-ready.
  • Committed main at f8bdbd3 is the source boundary; current uncommitted anatomy work is excluded and may only appear as future work.

Deliberate boundaries

  • Diagnose, recommend treatment, or accept a medical report.
  • Use three-dimensional polish to imply a finished medical product.
planned

Medical review for the MTC entry

Have the complete entry reviewed before changing its review-state label or making stronger quality claims.

planned

Source-backed directory expansion

Add topics only through the same structured source and acceptance path, with the denominator updated in the claim record.

investigating

Accessible spatial atlas

Evaluate whether a later 3D world improves understanding while preserving keyboard, reduced-motion, mobile, and no-canvas reading paths.

10 / Stack and sources

The systems underneath it, and where this account comes from.

  • TypeScript
  • structured medical content
  • 3D research
  • source metadata
Audited internallyOncoVerse committed repository, content directory, and MTC entry
Audited internallyPortfolio content-scope claim record