Sword strengthens AI Care model with new U.S. patent

From day one, we’ve believed that technology and AI would shape the future of care, and that innovation would protect it. That belief has guided our long-term investment in intellectual property, helping ensure that every breakthrough we deliver to members is built on a strong, defensible, and scientifically grounded foundation.

To date, our IP portfolio spans 53 distinct inventions protected by more than 90 patents and patent applications across the U.S. and Europe. More than legal protection, that portfolio reflects how innovating is a strategic priority at Sword.

This month, we’re proud to mark another major milestone: We’ve been granted a core U.S. patent for our proprietary multi-dimensional reinforcement learning system - a foundational technology that strengthens the AI engine behind our AI Care Model.

A new patent for continuous clinical learning

U.S. Patent No. 12,573,494 - “Multi-Dimensional Reinforcement Learning in an AI Platform” protects the dual-signal learning architecture that powers our AI engine.

While many healthcare AI systems remain largely static after training, our patented system is designed to improve continuously over time. It does so by learning simultaneously from two critical signals:

• Clinician validation of AI-generated treatment recommendations
• Real-world patient outcomes

This dual-feedback approach enables our AI to adapt safely and intelligently, improving its recommendations as more care is delivered.

Built for real-world clinical settings

This technology represents a significant advance in how AI can be deployed in healthcare.

Our AI combines three integrated systems:

• an observation layer that captures real-time biomechanical, physiological, and behavioral data
• a reasoning layer powered by specialized machine learning models
• a communication layer that delivers personalized guidance and feedback to patients

Together, these systems create a continuous learning cycle. Our AI generates treatment recommendations, supervising clinicians review them, and the system then learns from both clinical approval decisions and patient outcomes to progressively refine future recommendations.

Safety by design

At the core of the system is a human-in-the-loop framework.

AI recommendations are always reviewed and validated by a licensed clinician before implementation. That design allows us to continuously improve our models while preserving a high standard of safety and clinical oversight.

This patent represents the future of healthcare and precisely what AI Care was built to deliver: a continuous learning loop that figures out what works best for each individual patient through careful experimentation, made inherently safe due to the assurance of a human in the loop."

Luís Ungaro, Senior Vice President of Applied AI

A stronger foundation for AI Care

For us, patents are not just legal milestones. They are proof of the work happening behind the scenes to make AI Care more reliable, more adaptive, and more useful in real clinical settings. Each one brings us closer to our mission of helping billions of people live free from pain, recover faster, and reach their full potential.