Estimation of Fine-Grained Foot Strike Patterns with Wearable Smartwatch Devices

Identification of footstrike pattern using accelerometry and machine learning

Recent reports have suggested that there may be a relationship between footstrike pattern and overuse injury incidence and type. With the recent increase in wearable sensors, it is important to identify paradigms where the footstrike pattern can be detected in real-time from minimal data. Machine learning was used to classify tibial acceleration data into three distinct footstrike patterns: rearfoot, midfoot, or forefoot. Tibial accelerometry data were collected during treadmill running from 58 participants who each ran with rearfoot, midfoot, and forefoot strike patterns. These data were used as inputs into an artificial neural network classifier. Models were created by using three distinct acceleration data sets, using the first 100%, 75%, and 40% of stance phase. All models were able to predict the footstrike pattern with up to 89.9% average accuracy. The highest error was associated with the identification of the midfoot versus forefoot strike pattern. This technique required no pre-selection of features or filtering of the data and may be easily incorporated into a wearable device to aid with real-time footstrike pattern detection.

Fine-Grained Motion Recognition in At-Home Fitness Monitoring with Smartwatch: A Comparative Analysis of Explainable Deep Neural Networks

The squat is a multi-joint exercise widely used for everyday at-home fitness. Focusing on the fine-grained classification of squat motions, we propose a smartwatch-based wearable system that can recognize subtle motion differences. For data collection, 52 participants were asked to perform one correct squat and five incorrect squats with three different arm postures (straight arm, crossed arm, and hands on waist). We utilized deep neural network-based models and adopted a conventional machine learning method (random forest) as a baseline. Experimental results revealed that the bidirectional GRU/LSTMs with an attention mechanism and the arm posture of hands on waist achieved the best test accuracy (F1-score) of 0.854 (0.856). High-dimensional embeddings in the latent space learned by attention-based models exhibit more clustered distributions than those by other DNN models, indicating that attention-based models learned features from the complex multivariate time-series motion signals more efficiently. To understand the underlying decision-making process of the machine-learning system, we analyzed the result of attention-based RNN models. The bidirectional GRU/LSTMs show a consistent pattern of attention for defined squat classes, but these models weigh the attention to the different kinematic events of the squat motion (e.g., descending and ascending). However, there was no significant difference found in classification performance.