regi-Twin

entire health data dashboard
analytics integrated with live digital wearables data and synced across all platofrms likehevy,healthifyme,googlefit
base line health score calculated from the health score of people of that age
3d digital twin that gives meaningful insights of the data that was once meaningless and fragmented
the capture engine that turn the mobile camera to capturengine trained with 2 million images to make the elite healthcare accessible to all

Inspiration: Somewhere, an Olympic athlete steps onto a $200,000 motion-capture rig, surrounded by biomechanists reading their every joint angle in real time. A few miles away, an elderly woman loses her balance on a staircase — no one saw the decline coming, because no one was measuring. A factory worker's back gives out after years of silent strain. A patient abandons their post-op physio sheet because there's no one to tell them if they're even doing it right. 1.71 billion people live with musculoskeletal disorders — the #1 cause of disability worldwide — and almost none of them have access to what elite athletes take for granted. We asked: what if the only hardware you needed was the phone already in your pocket?

What It Does: PhysioTwin turns any smartphone camera into a clinical-grade biomechanics lab. Stand in front of it for 30 seconds — a sit-to-stand, a squat, a simple movement test — and it extracts 33 3D skeletal landmarks, no wearable required. That data builds a living 3D Digital Twin, a glowing anatomical model that lights up exactly which joints are under strain. Behind it, a 6-axis Capability Profile (mobility, stability, movement quality, cardio, recovery, reserve) tracks how your body changes over time. Ask it questions and it talks back — a conversational twin grounded in your own movement history. Tell it you're planning a run, and its predictive engine simulates the toll before you take a single step. It even reads MRI reports and translates medical jargon into plain guidance. All of this, on a $50 phone.

How We Built It: We made a deliberate bet: zero hardware, zero compromise. MediaPipe pose estimation runs entirely on-device via WebAssembly at 30+ FPS — no video ever leaves the phone, only sanitized joint coordinates. Those coordinates feed a mathematical biomechanical engine computing flexion, asymmetry, and angular velocity, with CUSUM change-point detection catching sub-clinical deterioration before it becomes a symptom. For reasoning, we paired FastAPI with Groq-hosted Llama 3.1 for sub-second clinical inference, and gave the twin a body using Three.js — a WebGL anatomical model with real-time strain heatmaps. The whole thing ships as an installable PWA that runs on low-bandwidth connections, because equity was the point from day one.

Challenges We Ran Into: Going wearable-free meant fighting for every ounce of precision from vision alone — matching lab-grade signal with nothing but a camera. Explainability was its own mountain: a joint-risk score means nothing if a user can't understand why the system is worried. And every design decision ran into the same wall — how do you give someone real, actionable insight without ever crossing the line into diagnosis?

Accomplishments We're Proud Of: We built a system that needs literally nothing but a phone camera to produce clinical-grade movement analysis. We're proud that our privacy model isn't a footnote — it's architectural: video never leaves the device, full stop. And we're proud that one engine flexes from an elite athlete optimizing performance to a grandmother trying not to fall, without feeling like two different products.

What We Learned: We learned a phone camera can carry far more biomechanical truth than we assumed going in. We learned that in health-adjacent tools, trust isn't a nice-to-have — it's the whole product, and it lives or dies on whether the system can explain itself. And we learned that designing for radically different users at once — athlete, elder, rural patient, clinician — doesn't dilute a product. It forces every feature to earn its place.

What's Next for PhysioTwin: Contactless heart rate and HRV estimation straight from camera (rPPG), full gait analysis, and multilingual voice guidance in the next few months. After that: clinical trials with orthopedic institutions, EHR/FHIR interoperability, and a B2B rollout that puts a diagnostic-grade dashboard in the hands of physiotherapists managing 100+ remote patients at once. The end goal hasn't changed since day one — healthcare isn't a luxury, it's a right, and the future of physical medicine should already be sitting in everyone's pocket.