Predicting and Detecting Freezing of Gait in Parkinson's Disease
Author Information
Author(s): Pardoel Scott, AlAkhras Ayham, Jafari Ensieh, Kofman Jonathan, Lemaire Edward D., Nantel Julie
Primary Institution: University of Waterloo
Hypothesis
Can machine learning improve the prediction and detection of freezing of gait episodes in individuals with Parkinson's disease using plantar-pressure data?
Conclusion
The study found that a machine learning model trained on a larger dataset significantly improved the prediction and detection of freezing of gait episodes in Parkinson's disease.
Supporting Evidence
- The model trained on Dataset 3 identified 86.84% of total freezing episodes.
- The model using Dataset 3 predicted freezing episodes 0.94 seconds before onset.
- Dataset 2 had the highest sensitivity at 82.50% for detecting freezing episodes.
- Removing non-freezers from Dataset 1 improved model sensitivity.
- Participants were tested under controlled conditions to increase the likelihood of freezing.
Takeaway
This study shows that a computer can help predict when people with Parkinson's might freeze while walking, which can help them avoid falling.
Methodology
The study used machine learning to analyze plantar-pressure data from participants with Parkinson's disease to predict and detect freezing of gait episodes.
Potential Biases
The presence of participants who did not experience freezing may have biased the model's performance.
Limitations
The study had a limited sample size and did not account for all variations of freezing of gait presentations.
Participant Demographics
Participants included 17 individuals with Parkinson's disease, with varying ages and disease durations.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website