Hidden Markov Model-Based Smart Annotation for Benchmark Cyclic Activity Recognition Database Using Wearables.

Activity monitoring using wearables is becoming ubiquitous, although accurate cycle level analysis, such as step-counting and gait analysis, are limited by a lack of realistic and labeled datasets. The effort required to obtain and annotate such datasets is massive, therefore we propose a smart annotation pipeline which reduces the number of events needing manual adjustment to 14%. For scenarios dominated by walking, this annotation effort is as low as 8%. The pipeline consists of three smart annotation approaches, namely edge detection of the pressure data, local cyclicity estimation, and iteratively trained hierarchical hidden Markov models. Using this pipeline, we have collected and labeled a dataset with over 150,000 labeled cycles, each with 2 phases, from 80 subjects, which we have made publicly available. The dataset consists of 12 different task-driven activities, 10 of which are cyclic. These activities include not only straight and steady-state motions, but also transitions, different ranges of bouts, and changing directions. Each participant wore 5 synchronized inertial measurement units (IMUs) on the wrists, shoes, and in a pocket, as well as pressure insoles and video. We believe that this dataset and smart annotation pipeline are a good basis for creating a benchmark dataset for validation of other semi- and unsupervised algorithms.

Robust Stride Segmentation of Inertial Signals Based on Local Cyclicity Estimation.

A novel approach for stride segmentation, gait sequence extraction, and gait event detection for inertial signals is presented. The approach operates by combining different local cyclicity estimators and sensor channels, and can additionally employ a priori knowledge on the fiducial points of gait events. The approach is universal as it can work on signals acquired by different inertial measurement unit (IMU) sensor types, is template-free, and operates unsupervised. A thorough evaluation was performed with two datasets: our own collected FRIgait dataset available for open use, containing long-term inertial measurements collected from 57 subjects using smartphones within the span of more than one year, and an FAU eGait dataset containing inertial data from shoe-mounted sensors collected from three cohorts of subjects: healthy, geriatric, and Parkinson’s disease patients. The evaluation was performed in controlled and uncontrolled conditions. When compared to the ground truth of the labelled FRIgait and eGait datasets, the results of our evaluation revealed the high robustness, efficiency (F-measure of about 98%), and accuracy (mean absolute error MAE in about the range of one sample) of the proposed approach. Based on these results, we conclude that the proposed approach shows great potential for its applicability in procedures and algorithms for movement analysis.

Inertial Sensor-Based Gait Recognition: A Review.

With the recent development of microelectromechanical systems (MEMS), inertial sensors have become widely used in the research of wearable gait analysis due to several factors, such as being easy-to-use and low-cost. Considering the fact that each individual has a unique way of walking, inertial sensors can be applied to the problem of gait recognition where assessed gait can be interpreted as a biometric trait. Thus, inertial sensor-based gait recognition has a great potential to play an important role in many security-related applications. Since inertial sensors are included in smart devices that are nowadays present at every step, inertial sensor-based gait recognition has become very attractive and emerging field of research that has provided many interesting discoveries recently. This paper provides a thorough and systematic review of current state-of-the-art in this field of research. Review procedure has revealed that the latest advanced inertial sensor-based gait recognition approaches are able to sufficiently recognise the users when relying on inertial data obtained during gait by single commercially available smart device in controlled circumstances, including fixed placement and small variations in gait. Furthermore, these approaches have also revealed considerable breakthrough by realistic use in uncontrolled circumstances, showing great potential for their further development and wide applicability.

Optimization of heartbeat detection in fiber-optic unobtrusive measurements by using maximum a posteriori probability estimation.

This paper deals with an optimized heartbeat detection multimethod by using maximum a posteriori probability (MAP). The approach was derived for unobtrusive fiber-optic measurements of cardiac activity. Multiple independent detection methods were selected and characterized by their delays and variability referring to cardiac electrical excitation (R waves). Validation of the approach was performed in two experiments involving 24 participants: 10 interferometric signals were recorded at rest, 14 with variable heart rate after physical exercise. The proposed MAP heartbeat detection was assessed by a cross validation in 250 iterations. Obtained results show the overall efficiency, which was estimated by a product of the sensitivity, precision, and variability of heartbeat detections, yields 97.04 ± 3.36% for the experiment with physical exercise and 97.07 ± 4.49% at rest. The method's accuracy guarantees that the heartbeat detections differ for 22 ± 5 ms and 22 ± 3 ms from the ECG reference in the two types of experiments, respectively.

Detection of heartbeat and respiration from optical interferometric signal by using wavelet transform.

A novel approach for the heartbeat and respiration detection based on optical interferometer and wavelet transform is proposed in this paper. Optical interferometer is a sensitive device that detects physical elongation of optical fibre due to external perturbations. Mechanical activity of cardiac muscle and respiration reflects in interferometric signal when the interferometer is in contact with human body and, thus, enables unobtrusive detection of human vital signs. The efficiency and accuracy of the proposed approach was estimated in two experimental protocols. The first one collected interferometric signals from 20 subjects during rest. In the second experiment, 10 participants cycled an ergometer until reaching their submaximal heart rate, and were measured immediately after that. Heartbeat detection results show high efficiency (99.46±1.11% sensitivity, 99.60±1.05% precision) and accuracy (mean relative error (MRE) of beat-to-beat intervals 3.16±2.32%) for the first experiment, and slightly lower efficiency (96.22±2.96% sensitivity, 95.35±3.03% precision) and accuracy (MRE of 9.56±3.67%) for the second experiment. Considering respiration detection, high efficiency (97.64±7.28% sensitivity, 99.38±2.80% precision) and accuracy (MRE of intervals between respiration events 7.37±7.20%) for the first experiment, and acceptable efficiency (92.05±6.10% sensitivity, 93.45±3.08% precision) and accuracy (MRE of 16.28±6.25%) for the second experiment confirm a practical value of proposed approaches.

Heartbeat and respiration detection from optical interferometric signals by using a multimethod approach.

In this paper, a multimethod approach for heartbeat and respiration detection from an optical interferometric signal is proposed. Optical interferometer is a sensitive device that detects physical changes of optical-fiber length due to external perturbations. When in direct or indirect contact with human body (e.g., hidden in a bed mattress), mechanical and acoustic activity of cardiac muscle and respiration reflect in the interferometric signal, enabling entirely unobtrusive monitoring of heartbeat and respiration. A novel, two-phased multimethod approach was developed for this purpose. The first phase selects best performing combinations of detection methods on a training set of signals. The second phase applies the selected methods to test set of signals and fuses all the detections of vital signs. The test set consisted of 14 subjects cycling an ergometer until reaching their submaximal heart rate. The following resting periods were analyzed showing high efficiency (98.18 ± 1.40% sensitivity and 97.04 ± 4.95% precision) and accuracy (mean absolute error of beat-to-beat intervals 22±9 ms) for heartbeat detection, and acceptable efficiency (90.06 ± 7.49% sensitivity and 94.21 ± 3.70% precision) and accuracy (mean absolute error of intervals between respiration events 0.33 ± 0.14 s) for respiration detection.

Optimization of heartbeat detection based on clustering and multimethod approach.

In this paper, an approach for optimizing heartbeat locations as detected in time by multimethod approach is proposed. The approach builds a two-dimensional representation of heartbeat locations obtained by several independent detection methods. The representation depends on heartbeat time instants and beat-to-beat intervals. It is first transformed into a smoothed two-dimensional histogram of points indicating individual heartbeat detections. Heartbeat time instants are determined as local maxima in the histogram. We tested our approach on signals acquired by optical interferometer. Seven subjects participated in the experiment, beginning by an ergometer exercise until they reached submaximal heart rate. A resting period followed, during which optical interferometric signal was taken unobtrusively in parallel with referential ECG. The proposed detection procedure was capable of tracking the changing heart rhythm by analyzing optical interferometric signals and comparing the results to the referential ECG recordings. Sensitivity 97.13±2.00% and precision 97.82±2.09% were obtained. Mean absolute error between detected beat-to-beat and referential RR intervals yielded 20.05±8.38 ms and corresponding mean relative error 7.47±3.19%.