Research Article
Preimpact Fall Detection for Elderly Based on Fractional Domain
Table 17
Comparison of the preimpact fall detection effect between fractional domain + multilayer dichotomy and other algorithms.
| Classification algorithm | Average lead time (walk/run) () | Number of used features | Average accuracy (walk/run) | Fall segmentation | Unable to detect the state of fall |
| Threshold method (single person, independent threshold) | 431/184 | 1 | | Yes | No | Threshold method (single person, unified threshold) | 233/156 | 1 | | No | No | Threshold method (multiple people, unified threshold) | 270/244 | 1 | | No | Yes | KNN | 427/311 | 20 | 0.85/0.61 | Yes | No | Bagged trees | 428/333 | 20 | 0.77/0.85 | Yes | No | Fractional domain + multilayer dichotomy | 456/240 | 2-3 | 0.9/0.87 | Yes | No |
|
|