Mumbai: Doctors in Mumbai are developing an advanced artificial intelligence (AI)-based system that can predict freezing of gait (FOG) in patients with Parkinson’s disease, a severe movement disorder that causes sudden inability to walk despite the intention to move.
The initiative is being led by specialists at Jaslok Hospital and Research Centre, in collaboration with the Paris Brain Institute, and aims to create a low-cost, scalable tool for early detection and better disease management.
About the Innovation
The AI tool uses computer vision and machine learning to analyse simple walking videos of patients, which can even be recorded using a smartphone.
Instead of relying on expensive hospital equipment, the system studies:
Body posture
Walking rhythm
Balance and movement patterns
Subtle changes in gait over time
The AI model then predicts:
Whether a patient is at risk of developing gait freezing
When such episodes are likely to begin
How the condition may progress
What is Freezing of Gait?
Freezing of gait is one of the most disabling symptoms of Parkinson’s disease, where patients suddenly feel as if their feet are “stuck to the ground.”
Medical studies show:
It affects up to 75–80% of advanced Parkinson’s patients
It leads to frequent falls, injuries, anxiety, and loss of independence
Why this AI Tool is Important
Currently, predicting gait freezing requires:
High-end gait analysis laboratories
Wearable motion sensors
Continuous clinical monitoring
The new AI approach removes these barriers by using simple video-based analysis, making early detection more affordable and accessible.
Doctors say this could help:
Identify high-risk patients earlier
Reduce fall-related injuries
Improve treatment planning
Support long-term mobility care
Expert Insight
Senior neurosurgeon Dr. Paresh Doshi, who is leading the project, said Parkinson’s disease is increasing in India and early prediction of motor decline is crucial.
He noted that freezing of gait severely impacts quality of life, often leading to social isolation, anxiety, and depression in patients.
Research Plan
The project will be developed in phases:
Training AI using historical patient walking videos
Studying data from hundreds of past cases
Testing the model on new patients for real-world validation
The final goal is to build an open-access, clinically usable diagnostic tool that hospitals can adopt easily.
Global Context
Similar AI-based research worldwide has shown that machine learning models can predict gait abnormalities with accuracy ranging from 78% to over 90%, depending on data quality and method used.
Experts say AI-powered gait analysis is emerging as a major tool in neurological disease prediction and monitoring.
