XAI-based gait analysis of patients walking with Knee-Ankle-Foot orthosis using video cameras

Explainability improves medical trust: The system provides explanations for its gait analysis predictions rather than just a confidence score, addressing a key concern medical professionals have about AI reliability in clinical settings.

AI can work in real-world home environments: By handling camera movements through super-resolution and pose estimation techniques, this system removes constraints that previously required static cameras and fixed distances—making gait analysis more practical for patients monitoring themselves at home.

Specific, measurable features drive decisions: The AI identifies seven clinically validated features (stride length, step duration, cadence, speed, etc.) that were statistically proven to differentiate between different orthosis types, grounding the system's recommendations in established medical science rather than opaque pattern-matching.

Smartphone-based monitoring democratizes access: Using portable devices like cell phones for gait analysis could make this technology accessible to more patients for ongoing monitoring, rather than limiting it to specialized clinical facilities.

Responsible AI requires both accuracy and interpretability: This work demonstrates that responsible AI in healthcare means not just achieving good predictions, but also explaining why those predictions are made in ways that medical professionals can understand and verify.