Clinical Human Gait Classification: Extreme Learning Machine Approach

Patil, Prithvi; Kumar, Kunal; Gaud, Neha; Semwal, Vijay Bhaskar

doi:10.1109/icasert.2019.8934463

Cited by 78 publications

(32 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In future work, we also plan to use smartphone sensor data to examine a patient’s gait patterns, in addition to performing her/his activity recognition. For example, there is existing work about analyzing the human gait using sensor data, including estimating gait parameters (e.g., average stride duration) as well as detecting abnormalities in gait [ 31 , 32 ]. In future studies, we can investigate methods to analyze the human gait using smartphone accelerometer and gyroscope data jointly.…”

Section: Discussionmentioning

confidence: 99%

Smartphone-Based Activity Recognition Using Multistream Movelets Combining Accelerometer and Gyroscope Data

Huang

Yan

Onnela

2022

Sensors

View full text Add to dashboard Cite

Physical activity patterns can reveal information about one’s health status. Built-in sensors in a smartphone, in comparison to a patient’s self-report, can collect activity recognition data more objectively, unobtrusively, and continuously. A variety of data analysis approaches have been proposed in the literature. In this study, we applied the movelet method to classify the activities performed using smartphone accelerometer and gyroscope data, which measure a phone’s acceleration and angular velocity, respectively. The movelet method constructs a personalized dictionary for each participant using training data and classifies activities in new data with the dictionary. Our results show that this method has the advantages of being interpretable and transparent. A unique aspect of our movelet application involves extracting unique information, optimally, from multiple sensors. In comparison to single-sensor applications, our approach jointly incorporates the accelerometer and gyroscope sensors with the movelet method. Our findings show that combining data from the two sensors can result in more accurate activity recognition than using each sensor alone. In particular, the joint-sensor method reduces errors of the gyroscope-only method in differentiating between standing and sitting. It also reduces errors in the accelerometer-only method when classifying vigorous activities.

show abstract

Section: Discussionmentioning

confidence: 99%

Smartphone-Based Activity Recognition Using Multistream Movelets Combining Accelerometer and Gyroscope Data

Huang

Yan

Onnela

2022

Sensors

View full text Add to dashboard Cite

show abstract

“…In contrast, when the person is out of the circle, the algorithm analyses the acceleration to identify an impending fall condition to trigger the protection system. It is important to know that an ADL activity has a stage that can be misinterpreted as a lack of balance, the swing phase [31,32]. The condition of the balanced boundary circle is not enough to reduce false alarms.…”

Section: Pre-impact Fall Detection Algorithmmentioning

confidence: 99%

Wearable Pre-Impact Fall Detection System Based on 3D Accelerometer and Subject’s Height

et al. 2022

View full text Add to dashboard Cite

This study presents a low-power wearable system able to predict a fall by detecting a pre-impact condition, performed through a simple analysis of motion data (acceleration) and height of the subject. The system can detect a fall in all directions with an average consumption of 5.91 mA; i.e., it can monitor the activity of daily living (ADL), whether or not a fall occurs. The entire detection system uses a single wearable tri-axis accelerometer placed on the waist for the comfort of the wearer during a long-term application. The algorithm is based on the following hypothesis: "A region defined as balanced boundary circle, based on the user's height, is characterized by the fact the chance that an actual fall happening is minimal. When an activity is classified outside this circle, an acceleration analysis is performed to determine an impending fall condition". Our threshold-based algorithm was validated experimentally, first with 9 young healthy volunteers performing both normal ADL and fall activities and then using 10 ADL and 5 falls from public SisFall dataset. The results show that falls could be detected with an average lead-time of 259 ms before the impact occurs, with minimal false alarms (97.7% specificity) and a sensitivity of 92.6%. This is a good lead-time achieved thus far in pre-impact fall detection, permitting the integration of our detection system in a wearable inflatable airbag for hip protection.

show abstract

“…Our paper addresses the methodology in general and does not focus on any particular use-case. The activities that are to be recognized can also be diverse, ranging from broad ones such as walking [ 10 ], which is the type addressed in our paper, to more specific ones, such as walking up- and downstairs and on slopes (e.g., [ 11 ]).…”

Section: Related Workmentioning

confidence: 99%

What Actually Works for Activity Recognition in Scenarios with Significant Domain Shift: Lessons Learned from the 2019 and 2020 Sussex-Huawei Challenges

Kalabakov

Stankoski

Kiprijanovska

et al. 2022

Sensors

View full text Add to dashboard Cite

From 2018 to 2021, the Sussex-Huawei Locomotion-Transportation Recognition Challenge presented different scenarios in which participants were tasked with recognizing eight different modes of locomotion and transportation using sensor data from smartphones. In 2019, the main challenge was using sensor data from one location to recognize activities with sensors in another location, while in the following year, the main challenge was using the sensor data of one person to recognize the activities of other persons. We use these two challenge scenarios as a framework in which to analyze the effectiveness of different components of a machine-learning pipeline for activity recognition. We show that: (i) selecting an appropriate (location-specific) portion of the available data for training can improve the F1 score by up to 10 percentage points (p. p.) compared to a more naive approach, (ii) separate models for human locomotion and for transportation in vehicles can yield an increase of roughly 1 p. p., (iii) using semi-supervised learning can, again, yield an increase of roughly 1 p. p., and (iv) temporal smoothing of predictions with Hidden Markov models, when applicable, can bring an improvement of almost 10 p. p. Our experiments also indicate that the usefulness of advanced feature selection techniques and clustering to create person-specific models is inconclusive and should be explored separately in each use-case.

show abstract

Clinical Human Gait Classification: Extreme Learning Machine Approach

Cited by 78 publications

References 18 publications

Smartphone-Based Activity Recognition Using Multistream Movelets Combining Accelerometer and Gyroscope Data

Smartphone-Based Activity Recognition Using Multistream Movelets Combining Accelerometer and Gyroscope Data

Wearable Pre-Impact Fall Detection System Based on 3D Accelerometer and Subject’s Height

What Actually Works for Activity Recognition in Scenarios with Significant Domain Shift: Lessons Learned from the 2019 and 2020 Sussex-Huawei Challenges

Contact Info

Product

Resources

About