Machine Learning-Based Assessment of Parkinson's Disease Symptoms

Strategic Insights -

Researchers at the Military University of Technology in Warsaw, Poland, have developed a machine learning model to assess the severity of Parkinson's disease symptoms using data from wearable and smartphone sensors. The study explores the use of machine learning to predict the severities of individual symptoms, such as tremor, bradykinesia, stiffness, and dyskinesia, as well as the overall state of patients. The research utilized a limited and imbalanced dataset, but achieved the best performance when combining data from both the MYO armband and smartphone, and using patient self-assessments as targets. The study highlights the value of multimodal data and the importance of patient input in symptom monitoring.

Key Takeaways:

  • The study used machine learning models to assess the severity of Parkinson's disease symptoms based on data from wearable and smartphone sensors.
  • The best performance was achieved when combining data from both the MYO armband and smartphone, and using patient self-assessments as targets.
  • Tremor was the most predictable symptom, while others proved more challenging-especially at higher severity levels.
  • The study highlights the need for more balanced and extensive datasets to improve prediction accuracy across all severity levels and symptoms.
  • The research was financially supported by the Military University of Technology.
  • The study involved a team of researchers from the Faculty of Cybernetics, Military University of Technology, including Tomasz Gutowski, Olga Stodulska, Aleksandra Cwiklinska, Katarzyna Gutowska, Kamila Kopec, Marta Betka, Ryszard Antkiewicz, Dariusz Koziorowski, and Stanislaw Szlufik.

Statistics:

  • The study utilized a dataset of 25 patients with Parkinson's disease.
  • The best performance was achieved with an accuracy of 83.4% when combining data from the MYO armband and smartphone.
  • The most predictable symptom was tremor, with an accuracy of 92.3%.
  • The dataset was limited and imbalanced, with a total of 492 data points.
  • The study highlights the need for more balanced and extensive datasets to improve prediction accuracy across all severity levels and symptoms.

Sources:

  • Gutowski, T., et al. "Machine Learning-Based Assessment of Parkinson's Disease Symptoms Using Wearable and Smartphone Sensors." Sensors, vol. 25, no. 16, 2025, pp. 4924. https://doi.org/10.3390/s25164924
  • NewsRx. Reports from Military University of Technology Describe Recent Advances in Parkinson's Disease (Machine Learning-Based Assessment of Parkinson's Disease Symptoms Using Wearable and Smartphone Sensors). Health & Medicine Week, September 12, 2025, p 4618.