New algorithm measures Parkinson’s tremor accurately in everyday life

For many people living with Parkinson’s disease, tremor, the characteristic shaking, is both bothersome and unpredictable. It may be almost absent during a consultation with the doctor, only to reappear at home while making coffee or sitting on the sofa. This makes it challenging to measure tremor as it actually occurs in daily life. Clinical examinations provide only a brief snapshot, and much of the variation throughout the day remains invisible. Wearable sensors (such as smartwatches) make continuous measurement possible, but earlier algorithms were often developed and tested in controlled settings. As a result, we lack robust methods that work regardless of context, whether the arm is at rest, walking, cooking, or using a computer mouse.

In a new study, researchers at SEFAS and Neuro-SysMed set out to measure tremor in the everyday lives of people with Parkinson’s disease. Using wearable sensors, they collected data from participants as they went about their normal daily routines. This allowed the team to develop a method for tracking tremor continuously throughout the day, without requiring the user to do anything other than wear the device.

An algorithm that listens for the smallest movements

The research team therefore developed the Tremor Index (TI), an algorithm that analyses movement data without relying on machine learning. Instead, it is based on signal processing techniques and wavelets, a mathematical tool that allows researchers to break down motion signals into different frequencies. This makes it possible to distinguish the small, rapid vibrations characteristic of tremor from the larger, slower movements we all perform throughout the day: swinging our arms, lifting objects, writing, cooking, or simply resting.

Tremor in Parkinson’s disease can occur somewhere between 3 and 12 Hz, and resting tremor, the most well-known form, often appears around 4–6 Hz. The algorithm examines these frequencies especially closely and attempts to identify individual patterns for each participant.

What the measurements revealed

Seven people with Parkinson’s disease took part in the study. Each participant wore the sensor for one week on one hand, and then for another week on the other. After the two weeks, the researchers compared the TI results from the more affected arm with the less affected, and evaluated these findings against a clinical assessment of tremor.

The patterns that emerged were clear: The hand that participants and clinicians described as the most affected by tremor also showed the highest TI in the sensor data. The differences were especially pronounced in the frequency ranges where resting tremor usually occurs, but higher frequencies associated with other tremor types were also visible. At the same time, the measurements showed that the algorithm was able to distinguish tremor from ordinary daily movements – something that is essential if measurements are to work in real-world conditions.

Another interesting finding was that the tremor frequencies themselves varied between participants. By tracking the TI over time, such patterns can reveal how tremor is influenced by medication, daily activities, and the severity of the disease.

A step towards a more precise follow-up

Although the study is small, the findings point towards a future in which tremor can be measured where it actually occurs – in daily life. With further testing in larger groups, the Tremor Index may become a valuable tool for both clinicians and researchers, offering a more detailed and deeper understanding of how the disease develops. For people with Parkinson’s, this may mean easier follow-up, less self-reporting, and a more precise assessment of treatment effects.

Read the article

You can read the article here “Wavelet-Based Tremor Quantification From Wrist-Worn Sensor Data in Home-Dwelling People With Parkinson's Disease (external link)” by Haakon Reithe, Monica Patrascu, Juan C. Torrado, Elise Førsund, Bettina S. Husebø, Simon U. Kverneng, Erika Sheard, Charalampos Tzoulis, and Brice Marty.