Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Machine Learning Strategies for Parkinson Tremor Classification Using Wearable Sensor Data

Jesus Paucar-Escalante, Matheus Alves da Silva, Bruno De Lima Sanches, Aurea Soriano-Vargas, Laura Silveira Moriyama, Esther Luna Colombini

TL;DR

This paper addresses the challenge of diagnosing and monitoring Parkinson's tremor using wearable sensor data and machine learning. It surveys data acquisition, preprocessing, feature extraction, model types, hyperparameter tuning, and validation strategies across 34 studies from 2018-2023, highlighting the shift from traditional ML to deep learning. It reveals wide heterogeneity in frequency bands, sensor configurations, and datasets, and notes a lack of external validation, underscoring the need for standardization and robust evaluation. The work informs researchers and clinicians about current capabilities and concrete directions to enable reliable, non-invasive tremor assessment in real-world settings.

Abstract

Parkinson's disease (PD) is a neurological disorder requiring early and accurate diagnosis for effective management. Machine learning (ML) has emerged as a powerful tool to enhance PD classification and diagnostic accuracy, particularly by leveraging wearable sensor data. This survey comprehensively reviews current ML methodologies used in classifying Parkinsonian tremors, evaluating various tremor data acquisition methodologies, signal preprocessing techniques, and feature selection methods across time and frequency domains, highlighting practical approaches for tremor classification. The survey explores ML models utilized in existing studies, ranging from traditional methods such as Support Vector Machines (SVM) and Random Forests to advanced deep learning architectures like Convolutional Neural Networks (CNN) and Long Short-Term Memory networks (LSTM). We assess the efficacy of these models in classifying tremor patterns associated with PD, considering their strengths and limitations. Furthermore, we discuss challenges and discrepancies in current research and broader challenges in applying ML to PD diagnosis using wearable sensor data. We also outline future research directions to advance ML applications in PD diagnostics, providing insights for researchers and practitioners.

Machine Learning Strategies for Parkinson Tremor Classification Using Wearable Sensor Data

TL;DR

This paper addresses the challenge of diagnosing and monitoring Parkinson's tremor using wearable sensor data and machine learning. It surveys data acquisition, preprocessing, feature extraction, model types, hyperparameter tuning, and validation strategies across 34 studies from 2018-2023, highlighting the shift from traditional ML to deep learning. It reveals wide heterogeneity in frequency bands, sensor configurations, and datasets, and notes a lack of external validation, underscoring the need for standardization and robust evaluation. The work informs researchers and clinicians about current capabilities and concrete directions to enable reliable, non-invasive tremor assessment in real-world settings.

Abstract

Parkinson's disease (PD) is a neurological disorder requiring early and accurate diagnosis for effective management. Machine learning (ML) has emerged as a powerful tool to enhance PD classification and diagnostic accuracy, particularly by leveraging wearable sensor data. This survey comprehensively reviews current ML methodologies used in classifying Parkinsonian tremors, evaluating various tremor data acquisition methodologies, signal preprocessing techniques, and feature selection methods across time and frequency domains, highlighting practical approaches for tremor classification. The survey explores ML models utilized in existing studies, ranging from traditional methods such as Support Vector Machines (SVM) and Random Forests to advanced deep learning architectures like Convolutional Neural Networks (CNN) and Long Short-Term Memory networks (LSTM). We assess the efficacy of these models in classifying tremor patterns associated with PD, considering their strengths and limitations. Furthermore, we discuss challenges and discrepancies in current research and broader challenges in applying ML to PD diagnosis using wearable sensor data. We also outline future research directions to advance ML applications in PD diagnostics, providing insights for researchers and practitioners.

Paper Structure

This paper contains 23 sections, 9 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Contributions
  3. Background
  4. Parkinson Disease
  5. Pathophysiology of Parkinson's disease
  6. Symptoms
  7. Motor symptoms
  8. Non-Motor symptoms
  9. Detection of tremors
  10. Taxonomy
  11. Studies Review
  12. Signal Adquisition
  13. Data Preprocessing
  14. Feature extraction
  15. Selection and Evaluation of AI models
  16. ...and 8 more sections

Figures (9)

  • Figure 1: Illustration of the proposed general structure (Taxonomy).
  • Figure 2: Number of studies per year.
  • Figure 3: Chronology illustrating studies aimed at Parkinson's classification using inertial sensors from 2018 to the end of 2023. The timeline includes a sub-classification dependent on the generation to which each architecture used in each study belongs, taking as a reference the classification of Skandlha et al. in skandha20203. The dates shown in the timeline are the reference dates of publication in their respective journals. The corresponding citation is in numerical form.
  • Figure 4: Distribution per generation.
  • Figure 5: Distribution of studies with a control group.
  • ...and 4 more figures