Related Experiment Video
Updated: Oct 9, 2025
09:37
Substantiating Appropriate Motion Capture Techniques for the Assessment of Nordic Walking Gait and Posture in Older Adults
Published on: May 12, 2016
8.9K
Constructing Prediction Models for Freezing of Gait by Nomogram and Machine Learning: A Longitudinal Study
Kun Xu1,2, Xiao-Xia Zhou2, Run-Cheng He2
1Collaborative Innovation Center for Brain Disorders, Laboratory of Brain Disorders, Ministry of Science and Technology, Beijing Institute of Brain Disorders, Capital Medical University, Beijing, China.
Frontiers in Neurology
|December 23, 2021
Summary
This study identified new risk factors for freezing of gait (FOG) in Parkinson's disease (PD), including less tremor, weight loss, and fatigue. Prediction models were developed to assess FOG risk in PD patients.
Area of Science:
- Neurology
- Clinical Medicine
- Biostatistics
Background:
- Freezing of gait (FOG) is a debilitating symptom in Parkinson's disease (PD).
- Predictive models for FOG development are limited, necessitating further research into associated clinical factors.
- Understanding these factors can aid in early identification and management of FOG in PD patients.
Purpose of the Study:
- To identify clinical measurements associated with FOG development in Chinese PD patients.
- To construct and validate prediction models for FOG using Cox regression and machine learning.
- To uncover novel risk factors for FOG in a longitudinal cohort.
Main Methods:
- A 1-year longitudinal study of 967 PD patients (Hoehn and Yahr stages 1-3) without baseline FOG.
- Data collection included clinical characteristics, medication, and standardized scales (UPDRS, PDQ-39, NMSS, etc.).
- Cox regression and random forests (RF) were used to build and validate prediction models on training and test sets.
Main Results:
- 26.4% of patients developed FOG during the study.
- Significant risk factors included longer disease duration, older age, higher H&Y stage, wearing-off, higher LEDD, and specific symptom scores (UPDRS, PDQ-39, NMSS, etc.).
- Novel risk factors identified: lower tremor symptom degree, unexplained weight loss, and higher fatigue. Prediction models showed good predictive ability (C-index 0.71 for nomogram, 0.74 for RF).
Conclusions:
- New risk factors for FOG in PD patients include reduced tremor, unexplained weight loss, and increased fatigue.
- Developed prediction models demonstrate acceptable accuracy for identifying patients at risk of FOG.
- These findings can inform clinical practice for early FOG detection and intervention in Parkinson's disease.
