Time-Series at the Edge: Tiny Separable CNNs for Wearable Gait Detection and Optimal Sensor Placement

We study on-device time-series analysis for gait detection in Parkinson’s disease (PD) from short windows of triaxial acceleration, targeting resource-constrained wearables and edge nodes. We compare magnitude thresholding to three 1D CNNs for time-series analysis: a literature baseline (separable convolutions) and two ultra-light models - one purely separable and one with residual connections. Using the BioStampRC21 dataset, 2 s windows at 30 Hz, and subject-independent leave-one-subject-out (LOSO) validation on 16 PwPD with chest-worn IMUs, our residual separable model (Model 2, 533 params) attains PR-AUC = 94.5%, F1 = 91.2%, MCC = 89.4%, matching or surpassing the baseline (5,552 params; PR-AUC = 93.7%, F1 = 90.5%, MCC = 88.5%) with approximately 10x fewer parameters. The smallest model (Model 1, 305 params) reaches PR-AUC = 94.0%, F1 = 91.0%, MCC = 89.1%. Thresholding obtains high recall (89.0%) but low precision (76.5%), yielding many false positives and high inter-subject variance. Sensor-position analysis (train-on-all) shows chest and thighs are most reliable; forearms degrade precision/recall due to non-gait arm motion; naive fusion of all sites does not outperform the best single site. Both compact CNNs execute within tight memory/latency budgets on STM32-class MCUs (sub-10 ms on low-power boards), enabling on-sensor gating of transmission/storage. Overall, ultra-light separable CNNs provide a superior accuracy-efficiency-generalization trade-off to fixed thresholds for wearable PD gait detection and underscore the value of tailored time-series models for edge deployment.

💡 Research Summary

This paper explores on-device time-series analysis for gait detection in Parkinson’s disease (PD) using short windows of triaxial acceleration data. The study targets resource-constrained wearables and edge nodes, comparing magnitude thresholding to three 1D CNNs: a literature baseline with separable convolutions, an ultra-light purely separable model, and a residual separable model. Using the BioStampRC21 dataset with 2-second windows at 30 Hz and subject-independent leave-one-subject-out (LOSO) validation on 16 PD patients wearing chest IMUs, the residual separable model (Model 2) achieves PR-AUC = 94.5%, F1 = 91.2%, MCC = 89.4% with only 533 parameters, outperforming or matching the baseline model (5,552 params; PR-AUC = 93.7%, F1 = 90.5%, MCC = 88.5%) while using approximately 10 times fewer parameters. The smallest model (Model 1) reaches PR-AUC = 94.0%, F1 = 91.0%, and MCC = 89.1% with just 305 parameters. Thresholding achieves high recall but low precision, leading to many false positives and high inter-subject variance. Sensor position analysis indicates that chest and thighs are the most reliable positions; forearms degrade precision/recall due to non-gait arm motion, and naive fusion of all sites does not outperform the best single site. Both compact CNNs execute within tight memory/latency budgets on STM32-class MCUs (sub-10 ms on low-power boards), enabling efficient on-sensor gating for transmission/storage. Overall, ultra-light separable CNNs provide a superior accuracy-efficiency-generalization trade-off to fixed thresholds for wearable PD gait detection and underscore the value of tailored time-series models for edge deployment.