Origin

Inspiration As a final-year physiotherapy student, I regularly observe how musculoskeletal (MSK) injuries occur not suddenly, but through cumulative micro-trauma caused by subtle, uncorrected movement errors. Joint valgus, abnormal load distribution, and asymmetry often remain hidden until acute pain or injury sets in. While specialized motion analysis laboratories provide objective biomechanical data, they are prohibitively expensive and inaccessible to everyday athletes, gym-goers, and rehab patients.

Partnering with Tanishka, a final-year electronics engineering student, we set out to bridge clinical physiotherapy with markerless computer vision. Our goal with Origin was clear: democratize biomechanical analysis by turning any smartphone camera into an objective, lab-grade injury risk assessment tool.

What It Does Origin is an AI-powered musculoskeletal health platform that detects movement deviations and predicts injury risks before they occur. Key functionalities include:

Markerless Motion Tracking: Uses standard mobile video feeds to track skeletal keypoints and joint angles in real time without hardware sensors.
3D Digital Twin & Biomechanical Modeling: Maps dynamic movements to evaluate joint stress, muscle activation, and kinetic chain symmetry.
Injury Risk Scoring: Computes joint-specific risk indices (such as ACL risk or patellofemoral overload) using dynamic angle calculations.
Real-Time Visual Biofeedback: Displays skeleton overlays, vector alignments, and heatmaps during movement tests to guide form.
Tele-Rehab Integration: Automatically generates standardized diagnostic reports to share with physiotherapists for remote monitoring.

How We Built It We combined clinical assessment logic with modern computer vision architectures:

Frontend & Camera Stream: Built a mobile interface that captures video frames at high frame rates and provides dynamic positioning overlays to guide user placement.
⁠Pose Detection Engine: Integrated markerless pose estimation algorithms to extract 2D/3D skeletal keypoint coordinates (xi,yi,zi ) for major joint centers.
Biomechanical Angle Calculations: Developed vector-based mathematical models to calculate dynamic joint angles continuously.
Risk Index & Recommendations: Algorithmic scoring models evaluate deviations against normative clinical thresholds, outputting visual heatmaps and tailored corrective exercise routines.

Challenges We Faced

Variable Camera Environments: Smartphone cameras operate across diverse lighting conditions, angles, and clothing types, making keypoint estimation noisy. We worked extensively on filtering pipeline algorithms to smooth angle telemetry.
Clinical vs. Algorithmic Precision: Standard goniometric assessments rely on manual palpation. Translating visual keypoints into clinically accurate biomechanical metrics required rigorous benchmarking to ensure variation remained below 3 degree .
Real-Time Edge Processing: Running keypoint detection, joint angle math, and biofeedback overlays simultaneously on mobile hardware without severe frame drops required strict model optimization.

What We Learned

Interdisciplinary Synergy: Fusing clinical physiotherapy protocols with electronics and software engineering opened up entirely new possibilities for accessible healthcare technology.
Edge-Based Data Privacy: We discovered that processing pose estimation directly on-device drastically improves user trust and ensures strict compliance with health data privacy frameworks.

What's Next for Origin

Clinical Pilot Expansion: Expanding pilot validations across partner physical therapy clinics and athletic centers.
Wearable Integration: Incorporating smart band sensor streams to complement visual motion capture with electromyography (EMG) or force data.
Enterprise Tele-Rehab: Deploying white-labeled SaaS tools for enterprise corporate wellness programs and remote patient monitoring (RPM).