Mumbai: City doctors are developing an artificial intelligence (AI)-based tool to predict whether and when a newly diagnosed Parkinson’s patient may develop freezing of gait, a common and highly disabling symptom that makes patients feel as if their feet are stuck to the ground.Jaslok Hospital, Pedder Road, has started work on the project ahead of World Parkinson’s Day on April 11. The tool, being developed in collaboration with the Paris Brain Institute in France, is expected to take about two years to complete. Freezing of gait affects up to 75%-80% of patients with advanced Parkinson’s disease. It can cause sudden difficulty in starting to walk, turning, or moving through narrow spaces, often leading to falls, injuries and loss of independence. Senior neurosurgeon Dr Paresh Doshi, who is leading the study, said Parkinson’s disease is increasing in India, including among younger patients. “There is no cure for Parkinson’s disease, but our tool will help patients estimate how rapidly their disease will progress and their treatment can be planned accordingly,” he said. The team chose to study freezing of gait because of the impact it has on a patient’s life. “Patients have reduced mobility, loss of independence that leads to social isolation. They also fear falling down and have a lot of anxiety and depression,” said the doctor. The AI system will analyse routine videos of patients walking, recorded even on a mobile phone, using computer vision and machine learning. It aims to detect subtle changes in gait and body balance to predict both the risk of freezing of gait and when it may begin. The tool will not require expensive gait analysers, which cost between Rs 80 lakh and Rs 1 crore, wearable devices or high-cost tests. As a part of the project, more than 400 videos of patients taken over the past 30 years will first be analysed using machine learning. “We will analyse the gait of these patients using over 35 points,” he said. The hospital’s neurosurgery department has clinical and video data of more than 750 patients. In the first phase, researchers will build the model using retrospective data from more than 150 patients who developed freezing of gait. In the second phase, it will be tested prospectively on 337 patients over up to three years. “The goal is to create an open-access, low-cost tool that helps clinicians identify high-risk patients earlier, plan treatment better and reduce falls and disability,” said the doctor.
