Wearable Sensor-Based IoT XAI Framework for Predicting Freezing of Gait in Parkinsons Disease

Early detection of a serious symptom is now possible: The research demonstrates that wearable sensors can predict Freezing of Gait (FOG)—a dangerous symptom where Parkinson's patients suddenly lose the ability to move—with up to 97% accuracy using XGBoost machine learning, potentially allowing for earlier intervention and treatment.

The AI system can explain its reasoning: Using SHAP analysis, the researchers made their AI model interpretable by identifying that gyroscope data (rotational movement) is the most important factor in FOG prediction, which helps doctors and patients understand why the system makes its predictions rather than treating it as a "black box."

Wearable technology enables continuous, real-world monitoring: The system uses affordable, accessible wearable sensors (ESP-32 microcontroller with LoRa communication) that patients can wear daily, moving beyond laboratory testing to practical, continuous health monitoring in people's actual lives.

Location tracking could provide immediate safety assistance: By combining FOG prediction with GPS tracking, the technology could alert caregivers or emergency services when a patient is at risk of falling, potentially preventing serious injuries and improving quality of life for Parkinson's patients.

This represents a practical step toward responsible AI in healthcare: The combination of high accuracy, model interpretability, and real-world applicability demonstrates how AI can be developed responsibly—being both powerful and understandable to the humans who depend on it.