Clinical validation
Tested where it is hard to test
FoGO Health has been evaluated in clinical settings and benchmarked against independent datasets, so the numbers hold up outside a single patient group.
Overall performance
0.00%
Accuracy
0.00%
AUC score
0.00
F1 score
0.00/min
False positives
Generalization
Validated across three independent datasets
A common failure in FoG detection is overfitting to one patient group. The model was validated across three distinct, independent datasets to confirm robust generalization.
DatasetF1 score
Daphnet0.00
CuPiD0.00
Turning0.00
Daphnet · CuPiD · Turning
Outperforming the baselines that matter
Proprietary feature extraction and an edge-optimized Random Forest hold their accuracy while staying on-device, where threshold methods lose accuracy and heavier deep-learning models cannot run at all.
| Method | F1 score | Latency | Edge capable |
|---|---|---|---|
| Freezing Index (threshold) | 0.68 | < 50 ms | Yes |
| Cloud-based CNN | 0.87 | > 800 ms | No |
| Ahilaya Edge AIOurs | 0.85 | 250 ms | Yes |