Activity Recognition in Individuals Walking With Assistive Devices: The Benefits of Device-Specific Models

Background Wearable sensors gather data that machine-learning models can convert into an identification of physical activities, a clinically relevant outcome measure. However, when individuals with disabilities upgrade to a new walking assistive device, their gait patterns can change, which could affect the accuracy of activity recognition. Objective The objective of this study was to assess whether we need to train an activity recognition model with labeled data from activities performed with the new assistive device, rather than data from the original device or from healthy individuals. Methods Data were collected from 11 healthy controls as well as from 11 age-matched individuals with disabilities who used a standard stance control knee-ankle-foot orthosis (KAFO), and then a computer-controlled adaptive KAFO (Ottobock C-Brace). All subjects performed a structured set of functional activities while wearing an accelerometer on their waist, and random forest classifiers were used as activity classification models. We examined both global models, which are trained on other subjects (healthy or disabled individuals), and personal models, which are trained and tested on the same subject. Results Median accuracies of global and personal models trained with data from the new KAFO were significantly higher (61% and 76%, respectively) than those of models that use data from the original KAFO (55% and 66%, respectively) (Wilcoxon signed-rank test, P=.006 and P=.01). These models also massively outperformed a global model trained on healthy subjects, which only achieved a median accuracy of 53%. Device-specific models conferred a major advantage for activity recognition. Conclusions Our results suggest that when patients use a new assistive device, labeled data from activities performed with the specific device are needed for maximal precision activity recognition. Personal device-specific models yield the highest accuracy in such scenarios, whereas models trained on healthy individuals perform poorly and should not be used in patient populations.

[1]  Jonathon W. Sensinger,et al.  Speed-Adaptation Mechanism: Robotic Prostheses Can Actively Regulate Joint Torque , 2014, IEEE Robotics & Automation Magazine.

[2]  S. Lord,et al.  A comparison of activity classification in younger and older cohorts using a smartphone , 2014, Physiological measurement.

[3]  Konrad P. Kording,et al.  Journal of Neuroscience Methods , 2013 .

[4]  Maayan Harel,et al.  Learn on Source, Refine on Target: A Model Transfer Learning Framework with Random Forests , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[5]  David C. Mohr,et al.  Making Activity Recognition Robust against Deceptive Behavior , 2015, PloS one.

[6]  Mitja Luštrek,et al.  Semi-supervised Learning for Adaptation of Human Activity Recognition Classifier to the User , 2011 .

[7]  Shyamal Patel,et al.  A review of wearable sensors and systems with application in rehabilitation , 2012, Journal of NeuroEngineering and Rehabilitation.

[8]  Guang-Zhong Yang,et al.  Sensor Positioning for Activity Recognition Using Wearable Accelerometers , 2011, IEEE Transactions on Biomedical Circuits and Systems.

[9]  Kenneth Meijer,et al.  Activity identification using body-mounted sensors—a review of classification techniques , 2009, Physiological measurement.

[10]  Jonathon W. Sensinger,et al.  Design and preliminary testing of the RIC hybrid knee prosthesis , 2015, 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC).

[11]  H. Dawes,et al.  Wearable accelerometry-based technology capable of assessing functional activities in neurological populations in community settings: a systematic review , 2014, Journal of NeuroEngineering and Rehabilitation.

[12]  Gary M. Weiss,et al.  The Impact of Personalization on Smartphone-Based Activity Recognition , 2012, AAAI 2012.

[13]  L. Piwek,et al.  The Rise of Consumer Health Wearables: Promises and Barriers , 2016, PLoS medicine.

[14]  Jin-Hyuk Hong,et al.  Toward Personalized Activity Recognition Systems With a Semipopulation Approach , 2016, IEEE Transactions on Human-Machine Systems.

[15]  Christian Poellabauer,et al.  Activity Recognition for Persons With Stroke Using Mobile Phone Technology: Toward Improved Performance in a Home Setting , 2017, Journal of medical Internet research.

[16]  Wanmin Wu,et al.  Classification Accuracies of Physical Activities Using Smartphone Motion Sensors , 2012, Journal of medical Internet research.

[17]  Angelo M. Sabatini,et al.  Machine Learning Methods for Classifying Human Physical Activity from On-Body Accelerometers , 2010, Sensors.

[18]  Ling Bao,et al.  Activity Recognition from User-Annotated Acceleration Data , 2004, Pervasive.

[19]  Friedrich Foerster,et al.  Detection of posture and motion by accelerometry : a validation study in ambulatory monitoring , 1999 .

[20]  Gearóid Ó Laighin,et al.  Comparing Supervised Learning Techniques on the Task of Physical Activity Recognition , 2013, IEEE Journal of Biomedical and Health Informatics.

[21]  Diane J. Cook,et al.  Transfer learning for activity recognition: a survey , 2013, Knowledge and Information Systems.

[22]  E. D. Lemaire,et al.  Evaluation of a smartphone human activity recognition application with able-bodied and stroke participants , 2016, Journal of NeuroEngineering and Rehabilitation.

[23]  Satoru Morita,et al.  Analysis of stroke patient walking dynamics using a tri-axial accelerometer. , 2009, Gait & posture.

[24]  Sean T. Green,et al.  Random forests for verbal autopsy analysis: multisite validation study using clinical diagnostic gold standards , 2011, Population health metrics.

[25]  Kent Larson,et al.  Real-Time Recognition of Physical Activities and Their Intensities Using Wireless Accelerometers and a Heart Rate Monitor , 2007, 2007 11th IEEE International Symposium on Wearable Computers.

[26]  Yiqiang Chen,et al.  Cross-People Mobile-Phone Based Activity Recognition , 2011, IJCAI.

[27]  Aakash Gupta,et al.  Activity recognition in patients with lower limb impairments: Do we need training data from each patient? , 2016, 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC).

[28]  Konrad Paul Kording,et al.  Using Mobile Phones for Activity Recognition in Parkinson’s Patients , 2012, Front. Neur..

[29]  Edward D. Lemaire,et al.  Feature Selection for Wearable Smartphone-Based Human Activity Recognition with Able bodied, Elderly, and Stroke Patients , 2015, PloS one.

[30]  Feng Tian,et al.  State of the Art Review of Knee–Ankle–Foot Orthoses , 2014, Annals of Biomedical Engineering.

[31]  Thomas Schmalz,et al.  A functional comparison of conventional knee–ankle–foot orthoses and a microprocessor-controlled leg orthosis system based on biomechanical parameters , 2016, Prosthetics and orthotics international.

[32]  Miguel A. Labrador,et al.  A Survey on Human Activity Recognition using Wearable Sensors , 2013, IEEE Communications Surveys & Tutorials.

[33]  Ye Xu,et al.  Enabling large-scale human activity inference on smartphones using community similarity networks (csn) , 2011, UbiComp '11.

[34]  B. Dobkin,et al.  Reliability and Validity of Bilateral Ankle Accelerometer Algorithms for Activity Recognition and Walking Speed After Stroke , 2011, Stroke.

[35]  Mohanraj Karunanithi,et al.  Review of accelerometry for determining daily activity among elderly patients. , 2011, Archives of physical medicine and rehabilitation.

[36]  Luigi Ferrucci,et al.  Assessing Daily Physical Activity in Older Adults: Unraveling the Complexity of Monitors, Measures, and Methods. , 2016, The journals of gerontology. Series A, Biological sciences and medical sciences.

[37]  Takuya Maekawa,et al.  Unsupervised Activity Recognition with User's Physical Characteristics Data , 2011, 2011 15th Annual International Symposium on Wearable Computers.