HackYeah 2026 ReviewViviOR
Kristin-Lina Todorova · kristin-lina-todorova/ViviOR_HackYeah26
Council score
Median of 3 models, weighted by the task's official criteriaA technically impressive and plausibly working virtual operating room that misses the Sport & Healthcare core of health data driving decisions and arrives without the required deck, visual proof of the running product, or AI-use disclosure.
A first-person virtual OR with physically simulated, cuttable tissue along Langer's lines and 17 scored Lichtenstein steps with an explained per-step debrief is ambitious and genuinely strong. C's 5 under-credits the idea itself; the discounts for existing surgical VR simulators and the missed task core are already taken elsewhere.
Members disagree here: scores range by 2.0 points.The council sides with C's middle view: the project is genuinely healthcare through surgical training and patient safety, but it serves none of sport, mental wellbeing or access to care, and no personal health data becomes a recommendation or plan. A's 5 over-credits that indirect link; B's 1 ignores the legitimate medical-education ground.
Members disagree here: scores range by 4.0 points.Browser delivery with no headset, a demo account, resume support and conservative clinical content make it genuinely usable for trainees. Usability stays unproven: mouse-driven surgery, a reported 38 MB WebGL download, no educator validation, and an empty how-to-open field.
The code shows real craft: a brand book with palette and fonts, HUD styles derived from it, a hyperlegible fallback and an ECG animation on the vitals monitor. But only 1 of 4 screenshots is described and it is a logo, so the running interface could not be verified.
7,924 lines of coherent Unity C# with a validated procedure engine, error penalties, an exam pass mark, a postMessage web bridge and a deployed build justify the high median. Zero tests, the missing deck and the doc-to-repo mismatch hold it back; B's 5 over-weights the unverified runtime.
Members disagree here: scores range by 2.0 points.Strengths
- Substantial, coherent C# implementing the claimed subsystems: XPBD tissue with cutting and gaping, a rope solver for the spermatic cord, procedural textures, vitals reacting to blood loss, equipment and team.
- A real decision output: every step is validated, errors are explained and penalised, exam mode logs silently with an 80 pass mark, and results are posted to the website over a documented protocol.
- A Unity WebGL build is actually deployed on Vercel with a demo account and a scripted demo, so delivery is real even though runtime behaviour is unverified.
- Clinical and licensing discipline: WHO time-out checklist, correct Lichtenstein anatomy and mesh margins, and a full third-party asset licence table.
Weaknesses
- No deck, the required round-one deliverable, and an almost empty submission form, so a reviewer sees little beyond the README.
- Missed task requirement: AI tool use is nowhere disclosed in the pack, while a repo doc reads as an AI agent task brief.
- Weak category fit: no sport, mental wellbeing or access-to-care angle, and no real health data driving a plan; the scores are trainee performance and the vitals are simulation.
- Zero tests and no visual evidence of the running simulator: only 1 of 4 screenshots is described and it is a logo, and the demo check confirms only an HTTP 200 page title.
Red flags
- May not be built for this task: it is a surgical training simulator, which is healthcare-adjacent, but it touches none of the task's core. There is no sport, physical health, mental wellbeing or access-to-care angle, and no real health data drives a decision; the scores are trainee performance and the vitals are simulated.
- The web-integration doc is written as an AI agent task brief and describes a Three.js simulator with 11 steps while the submitted repo is Unity with 17; with 7,924 LOC from one author in four commits over roughly four hours and no AI-use disclosure, the work appears to be undisclosed AI assistance.
claude:glm-5.3-flash61.580% agreedots-studio/dots-3-note-preview:free50.070% agreeinclusionai/ling-3.0-flash-sante:free50.080% agreeclaude:glm-5.3–judgeThe council, the evidence pack, the prompts and the queue are all on GitHub. If a review helped you, a star helps other teams find it.