aiqbench analysis
This research investigates the transferability of a CNN-BiLSTM deep learning model from hip-worn to wrist-worn accelerometers for sedentary behavior detection. The study demonstrates that while zero-shot performance degrades due to sensor placement shifts, fine-tuning offers consistent advantages over training transformer models from scratch on limited labeled data. These findings provide practical insights into leveraging pre-trained architectures for wearable health monitoring applications where labeling costs are high.
Sedentary Behavior Classification for Wearable Sensors with a CNN-BiLSTM Model
arXiv:2608.02946v1 Announce Type: new Abstract: Accurate detection of sedentary behavior is important for studying health risks related to prolonged sitting, but posture-based classification remains challenging with wearable sensors, especially at the wrist. We study whether a deep learning model trained on hip-worn accelerometer data can transfer to wrist-worn accelerometer data for sitting versus non-sitting classification. We use CHAP, a CNN-BiLSTM model originally developed for hip accelerometers, and evaluate its zero-shot performance on wrist data as well as its adaptation through finetuning with varying amounts of labeled wrist data. Experiments are conducted on the iWatch dataset with ground-truth posture labels derived from wearable cameras. The hip-trained model performs strongly on hip data without retraining, but accuracy drops on wrist data due to sensor placement shift. Finetuning CHAP provides consistent advantages over transformer models trained from scratch. These findings suggest that hip-based pretraining provides a useful starting point for wrist deployment, while highlighting the need for wrist-specific adaptation to handle higher signal variability.
Authors:Yuliang Chen, Weiwei Shi, Jingjing Zou, Rong Zablocki, Animesh Kumar, Jordan A. Carlson, Sheri J. Hartman, Mikael Anne Greenwood-Hickman, Paul R. Hibbing, Marta Jankowska, Jay Yang, Arun Kumar, Loki Natarajan
Abstract:Accurate detection of sedentary behavior is important for studying health risks related to prolonged sitting, but posture-based classification remains challenging with wearable sensors, especially at the wrist. We study whether a deep learning model trained on hip-worn accelerometer data can transfer to wrist-worn accelerometer data for sitting versus non-sitting classification. We use CHAP, a CNN-BiLSTM model originally developed for hip accelerometers, and evaluate its zero-shot performance on wrist data as well as its adaptation through finetuning with varying amounts of labeled wrist data. Experiments are conducted on the iWatch dataset with ground-truth posture labels derived from wearable cameras. The hip-trained model performs strongly on hip data without retraining, but accuracy drops on wrist data due to sensor placement shift. Finetuning CHAP provides consistent advantages over transformer models trained from scratch. These findings suggest that hip-based pretraining provides a useful starting point for wrist deployment, while highlighting the need for wrist-specific adaptation to handle higher signal variability.
Machine Learning (cs.LG)
arXiv:2608.02946 [cs.LG]
(or arXiv:2608.02946v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.02946 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 204 words
Related coverage
Learning Transferable Sensor Models via Language-Informed Pretraining
arXiv · Sep 30, 2026
SensorFM: Towards a general intelligence and interface for wearable health data
Google Research · Jul 9, 2026
Seeing beyond BMI: Estimating cardiometabolic risk with smartphone imagery
Google Research · Aug 17, 2026
MC-TRCM: Observation-Aware Recursive Fusion for Incomplete Mobile and Wearable Mental-Health Feature Views
arXiv · Oct 9, 2026