TL;DR
This paper presents a novel system using on-shoe wearable sensors to analyze turning abnormalities in Parkinson's Disease (PD) patients during a standardized gait test. The system effectively quantifies turning deficits, correlating them with disease severity and motor impairment, offering a promising tool for objective gait assessment and patient monitoring.
What Problem Does This Paper Address?
Parkinson's Disease (PD) significantly impacts mobility, with gait disturbances being a hallmark symptom. While straight walking is often studied, turning is a particularly challenging and risky aspect of gait for PD patients, even in early disease stages where straight walking may appear normal. Current clinical assessments of turning are often subjective and lack the granularity to capture subtle, early-stage deficits. Existing wearable sensor studies for gait analysis have primarily focused on upper-extremity placement, which can miss crucial lower-extremity biomechanics relevant to turning. There's a need for a more objective, sensitive, and accessible method to quantify turning abnormalities in PD, bridging the gap between clinical observation and advanced gait analysis.
Methods
The researchers developed a system utilizing on-shoe wearable sensors (Shimmer 2R/3 IMUs) to capture tri-axial accelerometer and gyroscope data at 102.4 Hz. These sensors were mounted laterally on each shoe. To isolate turning from the overall gait sequence during a standardized 4x10m walk with 180° turns, they employed a semi-automatic stride segmentation approach. Turning strides were identified using a threshold of turning angles greater than 20°, with transition strides (those immediately before or after a turn) also included. Spatio-temporal parameters were then extracted, categorized into per-stride (stride time, path length, stride length, stride velocity, swing width) and global per-turn (number of strides, total duration) parameters. These parameters were statistically analyzed using one-way ANOVA and Bonferroni's post hoc test to compare PD patients (grouped by disease severity) with healthy controls.
Datasets
The study utilized data from 108 Parkinson's Disease patients and 42 age-matched healthy controls. The PD patients were recruited from the movement disorder outpatient center at the University Hospital Erlangen. The study population was characterized by age, sex, height, BMI, Hoehn and Yahr (H&Y) stage, UPDRS-III total score, and specific UPDRS-III sub-items for gait and postural instability. Patients were selected based on specific criteria (H&Y < 4, UPDRS gait item < 3) and were assessed while "ON" medication.
Key Insights and Ideas
- Turning is a critical, early indicator of PD gait deficits: The study reinforces that turning abnormalities can be present even in mild PD, often before significant issues with straight walking are apparent.
- On-shoe sensors offer superior biomechanical resolution for turning: Placing sensors directly on the shoe provides more detailed information about foot and lower-extremity movements crucial for analyzing turning, compared to upper-extremity sensors.
- Global turning parameters are more sensitive than per-stride parameters for early detection: While both types of parameters showed differences, global measures like the number of strides and turning time were more effective in distinguishing between controls and early-stage PD patients.
- Objective turning analysis can complement clinical assessments: The system's ability to correlate turning parameters with established clinical scores (H&Y, UPDRS-III) suggests its potential as an objective tool for physicians.
- Turning analysis is vital for long-term monitoring and fall prevention: The clear turning deficits observed highlight the importance of incorporating turning analysis into continuous, out-of-clinic monitoring to inform preventative strategies for falls and freezing of gait.
Key Results
The study found significant differences in turning parameters between PD patients and controls, as well as across different stages of PD severity.
- Global Parameters: PD patients required significantly more time and a larger number of strides to complete a turn compared to controls. The "number of strides per turn" showed a particularly significant difference between controls and early-stage PD patients, as well as across disease stages.
- Per-Stride Parameters: Stride velocity, path length, stride length, and swing width showed significant differences between controls, mild, and severe PD stages across all clinical scores. Stride time, however, did not show clear changes.
- Correlation with Clinical Scores: The turning parameters demonstrated a strong correlation with clinical scores, especially the gait and postural instability sub-items of the UPDRS-III, which were more sensitive than the global H&Y and UPDRS-III scores.
- Effect Sizes: Global parameters, particularly "number of strides per turn" and "turning time," exhibited large effect sizes (η² values up to 0.232), indicating their high efficiency in differentiating groups. Path length also showed consistently higher effect sizes than stride length.
Limitations and Future Work
- Sensor Synchronization: A limitation noted is the lack of synchronized sensors between the left and right foot, which prevented the analysis of asymmetry in turning. Future work could involve synchronized sensors for a more comprehensive analysis.
- Preferred Turning Direction: The study did not account for the laterality of PD, as patients were instructed to turn in their preferred direction. Future research could investigate how turning to the more affected side impacts gait.
- Long-Term Monitoring Challenges: While the study demonstrated the feasibility of turning analysis in a controlled setting, adapting this to long-term, out-of-clinic monitoring presents challenges due to increased stride variability and diverse turning behaviors. Further research is needed to translate these findings to continuous monitoring systems.
- OFF Medication Assessment: The current study assessed patients "ON" medication. Future studies could explore turning analysis during "OFF" medication states to capture a broader range of motor fluctuations.
Summary updated