Machine Learning Models Classify Walker Posture to Prevent Falls in Older Adults
A recent study published on arXiv investigates the effectiveness of machine learning models in distinguishing between walker usage, sitting, standing, and various postures to enhance safety for seniors. The Geometric and XGBoost models showed the highest performance, with XGBoost achieving an impressive 99.84% accuracy in identifying walker selection and 99.69% for differentiating between sitting and standing. In posture classification, the Geometric method reached 89.9% accuracy across 8 postures, while XGBoost excelled with 99.24% accuracy for 17 postures. Additionally, deep learning models, including a 4-layer CNN and Encoder-Decoder CNN, surpassed 98% accuracy in binary classifications. This research underscores the potential of machine learning to improve fall prevention in smart walkers.
Key facts
- Falls among older adults are a significant public health concern.
- Study evaluates Geometric, XGBoost, SVM, and deep learning models.
- XGBoost achieved 99.84% accuracy for walker choice.
- XGBoost achieved 99.69% accuracy for standing vs. sitting.
- Geometric approach attained 89.9% accuracy for 8 postures.
- XGBoost obtained 99.24% accuracy for 17 postures.
- 4-layer CNN and Encoder-Decoder CNN had accuracies above 98%.
- Study underscores potential of machine learning to enhance safety.
Entities
Institutions
- arXiv