0
0
MUMBAI — In a significant technological leap for neurodegenerative care, doctors at Jaslok Hospital & Research Centre have launched a pioneering clinical study to develop an artificial intelligence (AI) tool capable of predicting “freezing of gait” (FoG) in Parkinson’s patients. The project, announced in coordination with World Parkinson’s Day, uses simple smartphone videos of a patient walking to forecast one of the most disabling and dangerous symptoms of the disease.
The initiative is a high-stakes collaboration between senior Indian neurosurgeon Dr. Paresh Doshi and the Paris Brain Institute in France. By leveraging machine learning to identify subtle motor changes invisible to the human eye, the team hopes to provide a low-cost, early-warning system that could prevent life-altering falls for millions of patients.
A Growing Crisis in Motion
Parkinson’s disease is the world’s second most common neurodegenerative disorder, and its footprint in India is expanding rapidly. Local experts warn that the country may soon house one of the largest populations of Parkinson’s patients globally.
“The rising numbers are putting immense pressure on neurology services,” says Dr. Doshi. He emphasizes that early motor changes and complications like freezing of gait are frequently under-recognized in routine clinical practice, leaving patients vulnerable to injury before interventions can begin.
How Video AI “Sees” the Future
The tool under development seeks to turn any smartphone into a sophisticated diagnostic laboratory. Using computer-vision techniques, the AI tracks more than 35 specific points on the human body as a patient walks. By analyzing balance, stride length, and turning mechanics, the algorithm estimates both the risk of a patient developing FoG and the likely timeframe for its onset.
To build this “predictive engine,” researchers are tapping into a massive goldmine of data: Jaslok’s neurosurgery database, which contains clinical and video records for over 750 patients spanning 30 years.
-
Phase One: Training the model on retrospective data from 150 patients who eventually developed freezing.
-
Phase Two: Prospectively testing predictions in 337 newly diagnosed patients, following them for up to three years to verify accuracy.
A major pillar of the project is affordability. Traditional gait-analysis labs require motion-capture systems costing between ₹80 lakh and ₹1 crore ($95,000–$120,000). By using standard video and open-access software, the Mumbai team aims to democratize high-tech care for rural and low-resource clinics.
The “Glued to the Floor” Phenomenon
Freezing of gait is often described by patients as a terrifying sensation where their feet feel “glued to the floor” while their upper body continues to move forward. This mismatch is a primary driver of falls and fractures.
Statistics highlight the severity of the issue:
-
Prevalence: Between 38% and 65% of Parkinson’s patients will experience FoG during their illness.
-
Fall Risk: Research indicates that 61% of all falls in Parkinson’s patients are related to freezing episodes.
-
Psychological Impact: In a cohort of 164 patients, over half of those with FoG reported a fall in the previous month, leading to significant anxiety, social isolation, and “fear of falling.”
“Once a patient experiences their first freeze, their world often shrinks,” says a movement disorder specialist. “They stop going out, which leads to muscle wasting and depression. Predicting this symptom before it starts allows us to intervene when the patient is still mobile and confident.”
Global Context: From Wearables to Video
The Mumbai initiative joins a global surge in digital health research. In 2024, a contest supported by the Michael J. Fox Foundation saw nearly 1,400 teams develop FoG-detection models using wearable sensors. While those models achieved high accuracy—often between 88% and 92%—they frequently require specialized hardware like accelerometers.
The Jaslok–Paris collaboration is unique because it removes the hardware barrier. By focusing on video footage, the tool could theoretically be used via tele-neurology, allowing a patient in a remote village to upload a video for assessment by a specialist in a major city.
Expert Perspectives and Clinical Reality
Independent specialists note that the project aligns with current clinical guidelines. A 2023 review published in the Journal of the American Geriatrics Society confirmed that gait speed and freezing patterns are the strongest predictors of future falls.
However, experts also urge a balanced view. While AI can process data points, freezing is influenced by complex factors including:
-
Cognitive load: Multitasking or stress often triggers a freeze.
-
Medication cycles: Levodopa levels significantly affect motor fluidity.
-
Environment: Narrow doorways or crowds are common triggers.
“Digital tools are decision-support aids; they don’t replace the neurologist,” notes the Parkinson’s Foundation. Once the AI flags a high-risk patient, clinicians can preemptively adjust medication, recommend specialized physiotherapy involving “cueing” (using sound or light to restart walking), and initiate home safety modifications.
Limitations and the Road Ahead
Despite the excitement, the Mumbai tool is still in the “validation” phase. It has not yet been cleared for general clinical use.
One primary challenge for video-based AI is “real-world noise.” Algorithms that work perfectly in a well-lit clinic often struggle with the shadows, cluttered hallways, or poor camera angles found in a patient’s home. Furthermore, models trained on specific populations may need “recalibration” to ensure they are accurate across different ethnicities and body types.
For now, the medical community remains cautiously optimistic. If the prospective study of 337 patients proves successful, this smartphone-based tool could become a standard part of Parkinson’s care, moving the needle from reactive treatment to proactive prevention.
Medical Disclaimer
Medical Disclaimer: This article is for informational purposes only and should not be considered medical advice. Always consult with qualified healthcare professionals before making any health-related decisions or changes to your treatment plan. The information presented here is based on current research and expert opinions, which may evolve as new evidence emerges.
References
- https://health.economictimes.indiatimes.com/news/industry/revolutionary-ai-tool-to-predict-freezing-of-gait-in-parkinsons-patients-developed-by-doctors-in-mumbai/130187012?utm_source=top_story&utm_medium=homepage
