Applying Wearable Sensors and Machine Learning to the Diagnostic Challenge of Distinguishing Parkinson’s Disease from Other Forms of Parkinsonism DOI Creative Commons
Rana Momtaz, Lisa Shulman, Ann L. Gruber‐Baldini

et al.

Biomedicines, Journal Year: 2025, Volume and Issue: 13(3), P. 572 - 572

Published: Feb. 25, 2025

Background/Objectives: Parkinson’s Disease (PD) and other forms of parkinsonism share motor symptoms, including tremor, bradykinesia, rigidity. The overlap in their clinical presentation creates a diagnostic challenge, as conventional methods rely heavily on expertise, which can be subjective inconsistent. This highlights the need for objective, data-driven approaches such machine learning (ML) this area. However, applying ML to datasets faces challenges imbalanced class distributions, small sample sizes non-PD parkinsonism, heterogeneity within group. Methods: study analyzed wearable sensor data from 260 PD participants 18 individuals with etiologically diverse were collected during mobility tasks using single placed lower back. We evaluated performance models distinguishing these two groups identified most informative classification. Additionally, we examined characteristics misclassified presented case studies common practice, uncertainty at patient’s initial visit changes diagnosis over time. also suggested potential steps address dataset limited models’ performance. Results: Feature importance analysis revealed Timed Up Go (TUG) task When TUG test alone, exceeded that combining all tasks, achieving balanced accuracy 78.2%, is 0.2% movement disorder experts. differences some scores between correctly falsely classified by our models. Conclusions: These findings demonstrate feasibility sensors differentiating parkinsonian disorders, addressing key its streamlining workflows.

Language: Английский

Applying Wearable Sensors and Machine Learning to the Diagnostic Challenge of Distinguishing Parkinson’s Disease from Other Forms of Parkinsonism DOI Creative Commons
Rana Momtaz, Lisa Shulman, Ann L. Gruber‐Baldini

et al.

Biomedicines, Journal Year: 2025, Volume and Issue: 13(3), P. 572 - 572

Published: Feb. 25, 2025

Background/Objectives: Parkinson’s Disease (PD) and other forms of parkinsonism share motor symptoms, including tremor, bradykinesia, rigidity. The overlap in their clinical presentation creates a diagnostic challenge, as conventional methods rely heavily on expertise, which can be subjective inconsistent. This highlights the need for objective, data-driven approaches such machine learning (ML) this area. However, applying ML to datasets faces challenges imbalanced class distributions, small sample sizes non-PD parkinsonism, heterogeneity within group. Methods: study analyzed wearable sensor data from 260 PD participants 18 individuals with etiologically diverse were collected during mobility tasks using single placed lower back. We evaluated performance models distinguishing these two groups identified most informative classification. Additionally, we examined characteristics misclassified presented case studies common practice, uncertainty at patient’s initial visit changes diagnosis over time. also suggested potential steps address dataset limited models’ performance. Results: Feature importance analysis revealed Timed Up Go (TUG) task When TUG test alone, exceeded that combining all tasks, achieving balanced accuracy 78.2%, is 0.2% movement disorder experts. differences some scores between correctly falsely classified by our models. Conclusions: These findings demonstrate feasibility sensors differentiating parkinsonian disorders, addressing key its streamlining workflows.

Language: Английский

Citations

0