Motion Analytics: A Case Study

By Nathan B. Smith

Motion analysis is becoming easier for the typical organization thanks to machine learning, which generates new use cases and applications that might provide value. To collect high-quality movement data, which enables motion analytics like those used to analyze elite athletes or employ sophisticated sensors that require a room full of equipment used to be necessary. However, things are beginning to shift. Complicated machine learning algorithms are beginning to harvest the information that complex technological instruments like lidar and specialized sensors previously offered to algorithms. Instead of needing a specific sensor vest or dedicated space, developers are discovering methods to correctly record things like yoga alignment using the camera inherent in contemporary smartphones (Lawton, 2020).

Discussion

Sports performance and pathologic gait are analyzed using biomechanical gait analysis. Motion capturing technologies, research methods, and data processing approaches have increased our understanding of gait biomechanics. Despite these developments, much biomechanical research over the past 20 years has examined injury risk variables in isolation. Multiple biomechanical and clinical variables likely interact and act as combined risk factors, so traditional biomechanical analysis methods (for example, analysis of discrete variables, such as peak angles, with a statistical hypothesis test, such as t-test or ANOVA) cannot capture the complexity of these relationships. Advanced multivariate analysis and machine learning approaches such as PCA and SVM have been utilized to discover complicated relationships (Phinyomark et al., 2018).

Most studies on the biomechanics of walking and running have employed kinematic data and concentrated on figuring out gait waveform events, including joint angles at touchdown, toe-off, mid-stance, and mid-swing. Descriptive information is frequently derived, including peak angles, excursion, and range of motion from the gait waveform. However, these conventional methods require a priori selection of traits, which depends on adequate background information and subjective judgment. As a result, a significant amount of the kinematic data are ignored even though they can include valuable information on the variations across groups. Modern data science techniques, in contrast, focus on the following key elements: initial input features; dimensionality reduction utilizing feature selection and feature extraction; and learning algorithms through classification and grouping. 

Technological improvements have given academics vast volumes of data to study for patterns. The 3D GAIT system collects 3D biomechanical gait data and transfers it to a central database. Traditional data analytics cannot manage these high data quantities; hence "big data" statistical approaches should be devised. All the strategies mentioned in this study show promise, inspire further work and illustrate the possibilities of employing data science in running gait biomechanics research (Phinyomark et al., 2018).

Gait analysis for athletes

A location-aware Google query for “3D Gait System” yielded a list of many local San Diego businesses that offer gait analysis for people buying running shoes. The most well know of which is Road Runner Sports. Gait analysis helps runners maintain healthy joints. Scientists suggest a twenty-minute video recording on a treadmill can alleviate runners' difficulties. Gait analysis identifies the cause of an injury or destructive behavior that may cause one. It begins with a strength and flexibility test that involves manipulation and exercises. A treadmill workout was captured on video. These exams are conducted at a few hospitals and sports performance institutes and span one to two hours, including a medical review. Patients get written pain-free evaluations of how to run (Futterman, 2014).

Gait analysis for disabled veterans

Leg prostheses have improved in design throughout time, but they are still incapable of actively adjusting to changing walking velocities in a way that a biological limb can. Compared to non-amputees, people with leg amputations who use commercially available passive-elastic prostheses use much more metabolic energy to walk at the same velocities, prefer to walk slower, and have incorrect biomechanics. A bionic prosthesis that mimics the function of a biological ankle during level-ground walking has been designed, mainly supplying the net positive effort necessary for various walking velocities. During level-ground walking, Herr and Grabowski (2012) compared the metabolic energy expenditures, preferred velocities, and biomechanical patterns of seven patients with a unilateral transtibial amputation using the bionic prosthesis and their passive-elastic prosthesis to those of seven non-amputees. Compared to a passive-elastic prosthesis, the bionic prosthesis reduced metabolic cost by 8%, increased trailing prosthetic leg mechanical work by 57%, decreased leading biological leg mechanical work by 10% across walking velocities of 0.75-1.75 m s1, and increased preferred walking velocity by 23%. Using a bionic prosthesis resulted in metabolic energy expenditure, preferred walking velocities, and biomechanical patterns that were not significantly different from those without an amputation (Herr & Grabowski, 2011).

OpenCV

Big data analytics (BDA) can now be performed by just about anyone with access to a personal computer. Python, R, and Julia are three of the most widely used programming languages that provide superb support for BDA. In the context of this discussion, one of the most popular Python libraries for video analytics is OpenCV. OpenCV is a tremendous open-source computer vision, machine learning, and image processing package. Now, it plays a significant role in real-time operation, which is very important in today's systems. Using it, one can process images and videos to identify objects, faces, or even the handwriting of a human. When integrated with other libraries, such as NumPy, python can process the OpenCV array structure for analysis. To Identify image patterns and their various features, computer vision engineers use vector space and perform mathematical operations on these features.

Conclusion

This discussion addressed big data analytics use and scope in motion analysis, emphasizing 3D gait analysis. Technology allows consumers to create data at any moment. The globe generates a lot of data every second. Video data is a significant contributor. In 3D gait analysis, a large portion of extensive data collection is video data acquired daily from biomechanical observation and test equipment. Such massive data needs a platform for storage, retrieval, processing, and analysis. Big data analytics facilitates this analysis. 

References

EMC Education Services. (2015). Data science and big data analytics: Discovering, analyzing, visualizing and presenting data (1st ed.). Hoboken, NJ: Wiley.

Futterman, M. (2014, September 2014). Gait Analysis: The serious runner's salvation. Retrieved from Elevate Physical Therapy & Fitness Website: https://elevateptfit.com/gait-analysis-2/#:~:text=Increasingly%20the%20runner%E2%80%99s%20road%20to%20healthy%20joints%20starts,a%20bad%20habit%20that%20may%20lead%20to%20one.

Herr, H. M., & Grabowski, A. M. (2011). Bionic Ankle-foot prosthesis normalizes walking gait for persons with leg amputations. Proceedings of the Royal Society of Biological Sciences, 279(1728), 457-464. https://doi.org/10.1098/rspb.2011.1194

Lawton, G. (2020, November 20). Understanding motion analytics, where It Is, and where it's going. EPAM Website: https://www.epam.com/about/newsroom/in-the-news/2020/understanding-motion-analytics-where-it-is-and-where-its-going

Phinyomark, A., Petri, G., Ibáñez-Marcelo, E., Osis, S. T., & Ferber, R. (2018). Analysis of big data in gait biomechanics: current trends and future directions. Journal of Medical and Biological Engineering, 38, 244-260. https://doi.org/10.1007/s40846-017-0297-2