Table 3.
A comparison of reported performance metrics in the literature for automatic meal detection and estimation algorithms with our proposed approach.
| Algorithm | Performance Metrics | |||||
|---|---|---|---|---|---|---|
| Precision | Recall | F-Score | Delay | Size Error | UQ | |
| Dassau et al. [16] | - | - | - | 30 min | - | ✗ |
| Ramkissoon et al. [17] | 92.5% | 82% | 0.87 | 38 min | - | ✗ |
| Samadi et al. [18] | 79% | 87% | 0.86 | - | 23% | ✗ |
| Samadi et al. [19] | 79% | 93.5% | 0.86 | 35 min | - | ✗ |
| Zheng et al. [23] | 93% | 88% | 0.91 | 26 min | - | ✗ |
| Xie and Wang [21] | 84% | 76% | 0.80 | 45 min | 43% | ✓ |
| Mahmoudi et al. [22] | - | 99.5% | - | 58 min | - | ✓ |
| Ours | 93% | 76% | 0.84 | 38 min | 31% | ✓ |
| Ours—Meals | 92% | 92% | 0.92 | 37 min | 19% | ✓ |
UQ = Uncertainty Quantification.