⚡ Quick Summary
This study developed a machine learning algorithm to accurately predict gait events using a single inertial measurement unit (IMU) that simulates smartphone usage. The results showed an impressive overall accuracy of 92% across various walking speeds and phone positions.
🔍 Key Details
- 📊 Participants: 52 healthy adults (35 males, 17 females)
- ⚙️ Technology: Long-short term memory neural network (LSTM-NN)
- 📏 Data Collection: IMU data segmented in 20-ms windows
- 🏆 Performance: Overall accuracy of 92%, with right-side predictions at ∼94%
🔑 Key Takeaways
- 📱 Smartphone Integration: The study mimicked real-world smartphone carrying scenarios.
- 💡 Innovative Approach: Utilized a single IMU for gait analysis, reducing fixation issues.
- 🏃♂️ High Accuracy: Achieved 92% accuracy in detecting gait events.
- ⏱️ Minimal Time Error: Median time error was less than 3% of the gait cycle duration (∼77 ms).
- 🌍 Potential for Remote Analysis: Opens avenues for smartphone-based remote gait analysis.
- 🔄 Need for Further Research: Future studies are essential to enhance generalizability.
📚 Background
The evaluation of human gait is crucial for improving mobility and assessing various health conditions. Traditional methods often require fixed sensors, which can be cumbersome and impractical in everyday settings. The advent of wearable technologies, particularly smartphones, presents an opportunity to streamline this process and make gait analysis more accessible.
🗒️ Study
Conducted with 52 healthy adults, the study aimed to create a machine learning algorithm that predicts gait events such as heel strikes and toe-offs using a single IMU. Participants walked on a treadmill while carrying a surrogate smartphone in different positions, allowing researchers to gather comprehensive data on gait dynamics.
📈 Results
The LSTM-NN model demonstrated an overall accuracy of 92% in classifying gait events, with right-side predictions slightly outperforming left-side predictions at approximately 94% and 91%, respectively. Additionally, the median time error was found to be less than 3% of the gait cycle duration, indicating a high level of precision in the predictions.
🌍 Impact and Implications
The findings from this study represent a significant advancement in the field of gait analysis. By utilizing smartphones for remote gait monitoring, healthcare professionals can potentially offer more convenient and effective assessments without the need for cumbersome sensor fixation. This could lead to improved patient outcomes and greater accessibility to gait analysis in various settings.
🔮 Conclusion
This research highlights the promising potential of integrating machine learning with smartphone technology for gait analysis. With an impressive accuracy of 92% and minimal time error, the study paves the way for future innovations in remote health monitoring. Continued research is essential to enhance the generalizability of these findings and explore practical applications in real-world scenarios.
💬 Your comments
What are your thoughts on the use of smartphones for gait analysis? We’d love to hear your opinions! 💬 Join the conversation in the comments below or connect with us on social media:
Accurate detection of gait events using neural networks and IMU data mimicking real-world smartphone usage.
Abstract
Wearable technologies such as inertial measurement units (IMUs) can be used to evaluate human gait and improve mobility, but sensor fixation is still a limitation that needs to be addressed. Therefore, aim of this study was to create a machine learning algorithm to predict gait events using a single IMU mimicking the carrying of a smartphone. Fifty-two healthy adults (35 males/17 females) walked on a treadmill at various speeds while carrying a surrogate smartphone in the right hand, front right trouser pocket, and right jacket pocket. Ground-truth gait events (e.g. heel strikes and toe-offs) were determined bilaterally using a gold standard optical motion capture system. The tri-dimensional accelerometer and gyroscope data were segmented in 20-ms windows, which were labelled as containing or not the gait events. A long-short term memory neural network (LSTM-NN) was used to classify the 20-ms windows as containing the heel strike or toe-off for the right or left legs, using 80% of the data for training and 20% of the data for testing. The results demonstrated an overall accuracy of 92% across all phone positions and walking speeds, with a slightly higher accuracy for the right-side predictions (∼94%) when compared to the left side (∼91%). Moreover, we found a median time error <3% of the gait cycle duration across all speeds and positions (∼77 ms). Our results represent a promising first step towards using smartphones for remote gait analysis without requiring IMU fixation, but further research is needed to enhance generalizability and explore real-world deployment.
Author: [‘Larsen AG’, ‘Sadolin LØ’, ‘Thomsen TR’, ‘Oliveira AS’]
Journal: Comput Methods Biomech Biomed Engin
Citation: Larsen AG, et al. Accurate detection of gait events using neural networks and IMU data mimicking real-world smartphone usage. Accurate detection of gait events using neural networks and IMU data mimicking real-world smartphone usage. 2024; (unknown volume):1-11. doi: 10.1080/10255842.2024.2423252