Related Experiment Videos
A data mining methodology for predicting early stage Parkinson's disease using non-invasive, high-dimensional gait
Conrad Tucker1,2, Yixiang Han3, Harriet Black Nembhard3
1Industrial and Manufacturing Engineering, Engineering Design, Computer Science and Engineering, The Pennsylvania State University, University Park, PA 16802, USA.
IIE Transactions on Healthcare Systems Engineering
|March 16, 2018
Summary
This study introduces a data mining approach using low-cost sensors to detect Parkinson's disease (PD) movement abnormalities. This method offers a non-invasive, accessible way for early detection and remote monitoring of PD.
Area of Science:
- Neurology
- Biomedical Engineering
- Data Science
Background:
- Parkinson's disease (PD) is a prevalent neurological disorder characterized by motor impairments.
- Current diagnostic methods like CT scans are costly and may involve risks.
- Remote patient assessment is limited by the reliance on in-person, qualitative evaluations.
Purpose of the Study:
- To develop and evaluate a data mining methodology for detecting Parkinson's disease (PD) using non-invasive sensors.
- To assess the ability of low-cost hardware and data mining algorithms to classify PD cases and controls.
- To enable remote monitoring of PD progression and facilitate earlier diagnosis.
Main Methods:
- Utilized a 10-fold cross-validation approach to compare various data mining algorithms on gait data.
- Employed low-cost, non-invasive sensors to collect movement, gait, and posture data.
- Quantified the predictive accuracy of the data mining model using unseen test data.
Main Results:
- The proposed data mining methodology effectively classifies Parkinson's disease (PD) cases and controls.
- The study identified data mining algorithms that provide consistent results across varying gait data.
- The model demonstrated predictive accuracy in identifying PD-related movement abnormalities.
Conclusions:
- Non-invasive, low-cost sensors combined with data mining offer a feasible approach for PD detection.
- This methodology supports remote monitoring of gait features, potentially leading to earlier diagnosis.
- The findings pave the way for accessible, widespread screening of neurological disorders.