Algorithm helps in the treatment of Parkinson's disease
Tool developed by physicist quantifies tremors captured by sensor
Algorithm helps in the treatment of Parkinson's disease
Tool developed by physicist quantifies tremors captured by sensor
Around 2% of the world's elderly suffer from Parkinson's disease, a degenerative disease whose symptoms include tremors and slow movements. One of the challenges posed by Parkinson's is the difficulty that medical teams have in monitoring patients' treatment, since it is up to the patient to identify the progression of their condition. With the aim of creating a mechanism to aid in the analysis of tremors and, consequently, in defining therapy and medication dosage, physicist Caetano Ternes Coimbra developed an algorithm capable of quantifying movements captured by a wearable sensor (a smart wristwatch), the result of his master's dissertation, defended at the Gleb Wataghin Physics Institute (IFGW) at Unicamp. The research is part of a larger project that integrates the Hub of Artificial Intelligence Applied to Health and Well-Being – Viva Bem.
Supervised by Professor Rickson Coelho Mesquita and co-supervised by Professor Gabriela Castellano, Coimbra began his research shortly after the creation, in 2022, of the Viva Bem Hub, which focuses on the use of smartwatches for monitoring health and well-being. The project is sponsored by Samsung Eletrônica da Amazônia Ltda, within the scope of the Information Program Technology Law 8.248/91. In total, ten applications are developed for a commercial watch for daily use – a Samsung Galaxy Watch –, offering a practical, low-cost and reliable solution for monitoring, for example, sleep, anxiety, hypertension, diabetes and pathological and physiological tremors, including those resulting from Parkinson's.
“The master’s degree is like a pilot,” says Coimbra. “Our goal was to measure tremors using a commercial watch that people use every day, in different applications. Based on the literature in the field, we already knew that, at some level, we would be able to measure movement. But the challenge was to measure accurately using the watch’s sensor and all the technology involved, capable of detecting information such as arm movement,” explains the scientist, who graduated from the Federal University of Santa Catarina (UFSC), where he studied particle physics, “a completely different field.”
In addition to capturing data, Coimbra sought to create a deterministic algorithm to determine under what conditions the tremor occurred. “This is a very important point. I wanted to do something more explainable. I wanted to understand the signal in order to quantify it.” The idea is not to diagnose the condition using a watch, especially since this is a multifactorial disease. The watch’s main contribution is to provide support for adjusting treatment. Patients take home a watch that will automatically and continuously measure their tremor, recording at what time it improved or worsened.
XNUMXst step
“For over a decade, our laboratory has been developing research aimed at collecting relevant physiological information for various clinical applications, with a special focus on neurological and vascular diseases. The development and availability of wearable sensors allows us to go beyond the controlled environment of the laboratory, enabling us to collect data at any time and place,” adds the advisor. “Ternes’ research represents a first step in the use of a specific type of sensor for a given disease. Our goal is to move forward, exploring the potential application of these sensors in other neurodegenerative and vascular diseases.”
During the study, clinical data were also collected. “The doctor needs to analyze the patient’s tremor. But how can he do this during the consultation? The doctor has to trust what the patient says, which is not always accurate or reliable, especially among the elderly,” the researcher points out. Among the more than 4 million patients with Parkinson’s disease, the vast majority are elderly. In only 10% of cases is a diagnosis made before the patient is 45 years old.
“I had never dealt directly with patients. I am a physicist in the medical field, doing tests,” notes Coimbra. “We tested the app on some patients who had great difficulty moving, a fragile population. I was not prepared to do this, but the team of doctors [from the Hospital de Clínicas – HC – at Unicamp] participated, because this is a multidisciplinary work.”
Medical team
Laura Silveira Moriyama, coordinator of the Parkinson’s and movement disorders service at HC and professor of neurology at the University’s School of Medical Sciences (FCM), led the medical team involved in the research, which she claims is an important contribution to the treatment of movement disorders. “There are several types of neurological problems that affect the way a person moves, how they speak, swallow or use their hands. As research advances, we have increasingly interdisciplinary collaboration from areas such as engineering, physics and computing. In this project, Ternes collected objective measurements of the patient’s tremor and difficulty in movement. We, doctors, see the tremor. He, as a physicist, sees the frequency and amplitude of the tremor,” said Moriyama.
According to the neurologist, it is common for patients to be unsure whether their condition has improved after taking the medication. “We tried to find ways to measure this objectively. The sensor gives us more objective information.” The protocol created by Coimbra involves two types of equipment: electroneuromyography (which measures the activation of the muscles responsible for the tremor) and the accelerometry watch (which measures the acceleration at which that part of the body moved).
“We use it on the wrist, because Parkinson’s tremors affect the arms a lot, but it could be used on other parts of the body, such as the legs, torso and head. This type of sensor can help us see the progression of the disease.” For the doctor, the visual representation created by Coimbra also helped the patient understand their medical condition. “The sensor provides security for the patient and the doctor. There is no way to send the doctor to the patient’s home, but it is as if the patient takes a little piece of our team with them.”
The algorithm
Coimbra developed a classic algorithm using Fourier analysis to extract the physically relevant characteristics of tremor (an oscillatory signal that comes and goes) – its amplitude, frequency and frequency variation. “The algorithm has the potential to be applied to other symptoms, such as bradykinesia [slowness of movement], in addition to tremors. Measuring tremor is more than measuring a pathology. Everyone has a physiological tremor, because there are several things that can cause tremor, such as anxiety during a job interview or even caffeine intake. So measuring tremor is something very broad and it is not a new idea”, said the researcher, adding that there are many studies in the literature on the subject and, in recent years, improvements have been made in the miniaturization of sensors. “Most pathological tremors cannot be cured. Therefore, the main focus has been on effective management of symptoms. I think it is cool to do applied physics and be able to see concrete results. This is something challenging.”