Hippos
Hippos is building the first Human Body Model, a machine-learning system that predicts body state transitions to prevent injuries before they occur. The system ships in a soft exoskeleton for the knee that senses signals at 6,400 Hz, runs edge inference in under 5 ms, and intervenes via software alerts or a pneumatic airbag that activates faster than human reflex. The company has deployed the system with partners including the U.S. Army and Cleveland Clinic.
- Healthcare Technology
- Medical Devices
- Artificial Intelligence
- Sports Technology
- Robotics
- Internet of Things
- Hardware
- Wellness & Fitness Technology
Funding
Founders
Founder details are not available yet.
Product
Problem
Current wearable technology can only react to injuries after they happen, leaving the body's warning signals unread until it is too late. No machine can predict what a person's body will do next, how much capacity remains today, or that a knee is moments from giving out—so injuries occur that could have been prevented with early intervention.
Solution
Hippos is building the first Human Body Model, a machine-learning system that learns the transition between body states: state, action, context, intervention, and next state. The model ships inside a closed-loop physical intelligence system whose first form factor is a soft exoskeleton for the knee. The system senses body signals at 6,400 Hz on proprietary infrastructure, runs edge inference in under 5 ms with 95.78% average accuracy in clinical validation, and intervenes through both software alerts and a pneumatic micro airbag that activates faster than human reflex. This loop enables injury prevention today, with a roadmap toward extending lifespan and eventually supporting human augmentation. Each intervention changes an outcome and each outcome teaches the model, creating a self-improving dataset across military, elite sport, and rehabilitation deployments.
Target Audience
Primary users are athletes, physically active individuals, and workers in physically demanding roles who face elevated injury risk and need real-time protection and load guidance. The company also targets militaries, sports teams, surgeons, and health systems as institutional buyers.
Features
- Proprietary multimodal sensing infrastructure sampling at 6,400 Hz for high-resolution body-state tracking
- Edge inference models with sub-5 ms latency and 95.78% average accuracy in clinical validation
- Pneumatic micro airbag hardware that deploys faster than human reflex to protect joints during high-risk movements
- Kinematic risk alerts that notify users of impending injury before physical failure occurs
- Prescriptive load planning software that recommends safe activity levels based on predicted body capacity
- A transition-learning model that captures state, action, context, and intervention data from every hour of wear
- Closed-loop latency under 30 ms demonstrated in cadaveric testing with a 0.15% false positive rate across 54 live deployments and 4,000 simulations
- Active intervention reduced peak valgus angle change from 6.1° to 1.95° in single-leg landing studies across 17 athletes