An Intelligent Optimization Strategy Based on Deep Reinforcement Learning for Step Counting

With the popularity of Internet of things technology and intelligent devices, the application prospect of accurate step counting has gained more and more attention. To solve the problems that the existing algorithms use threshold to filter noise, and the parameters cannot be updated in time, an intelligent optimization strategy based on deep reinforcement learning is proposed. In this study, the counting problem is transformed into a serialization decision optimization. This study integrates the noise recognition and the user feedback to update parameters. The end-to-end processing is direct, which alleviates the inaccuracy of step counting in the follow-up step counting module caused by the inaccuracy of noise filtering in the two-stage processing and makes the model parameters continuously updated. Finally, the experimental results show that the proposed model achieves superior performance to existing approaches.

Head Trajectory Diagrams for Gait Symmetry AnalysisUsing a Single Head-Worn IMU

Gait symmetry analysis plays an important role in the diagnosis and rehabilitation of pathological gait. Recently, wearable devices have also been developed for simple gait analysis solutions. However, measurement in clinical settings can differ from gait in daily life, and simple wearable devices are restricted to a few parameters, providing one-sided trajectories of one arm or leg. Therefore, head-worn devices with sensors (e.g., earbuds) should be considered to analyze gait symmetry because the head sways towards the left and right side depending on steps. This paper proposed new visualization methods using head-worn sensors, able to facilitate gait symmetry analysis outside as well as inside. Data were collected with an inertial measurement unit (IMU) based motion capture system when twelve participants walked on the 400-m running track. From head trajectories on the transverse and frontal plane, three types of diagrams were displayed, and five concepts of parameters were measured for gait symmetry analysis. The mean absolute percentage error (MAPE) of step counting was lower than 0.65%, representing the reliability of measured parameters. The methods enable also left-right step recognition (MAPE ≤ 2.13%). This study can support maintenance and relearning of a balanced healthy gait in various areas with simple and easy-to-use devices.

Smartwatch Step-Counting App for Older Adults: Development and Evaluation Study (Preprint)

BACKGROUND Older adults who engage in physical activity can reduce their risk of mobility and disability. Short amounts of walking can improve their quality of life, physical function, and cardiovascular health. Various programs have been implemented to encourage older adults to engage in physical activity, but sustaining their motivation continues to be a challenge. Ubiquitous devices, such as mobile phones and smartwatches, coupled with machine-learning algorithms, can potentially encourage older adults to be more physically active. Current algorithms that are deployed in consumer devices (e.g., Fitbit) are proprietary, often are not tailored to the movements of older adults and have been shown to be inaccurate in clinical settings. Few studies have developed step-counting algorithms for smartwatches – but only using data from younger adults and often validating them only in controlled laboratory settings. OBJECTIVE In this work, we sought to develop and validate a smartwatch step-counting app targeting older adults that has been evaluated in free-living settings over a long period of time (24 weeks) with a large sample (N=42). METHODS the steps of older adults. The app includes algorithms to infer the level of physical activity and to count steps. We validated the step-counting algorithm with a total of 42 older adults in the lab (counting from a video recording, N= 20) and in free-living conditions — one 2-day field study (N=6) and two 12-week field studies (using the Fitbit as ground truth, N=16). During system development, we evaluated four kinds of walking patterns: normal, fast, up and down a staircase, and intermittent speed. For the field study, we evaluated various values for algorithm parameters, and subsequently evaluated the method’s performance using correlations and error rates. RESULTS The results from the evaluation showed that our step-counting algorithm performs well, highly correlated with the ground truth and with low error rate. For the lab study, there was stronger correlation for normal walking R2=0.5; across all activities, the Amulet was on average 3.2 (2.1%) steps lower (SD = 25.9) than video-validated steps. For the 2-day field study, the best parameter settings led to an association between Amulet and Fitbit (R2 of 0.989) and 3.1% (SD=25.1) steps lower than Fitbit, respectively. For the 12-week field study, the best parameter setting led to an R2 of 0.669. CONCLUSIONS Our findings demonstrate the importance of an iterative process in algorithm development in advance of field-based deployment. This work highlights various challenges and insights involved in developing and validating monitoring systems in real-world settings. Nonetheless, our step-counting app for older adults had good performance relative to the ground truth (a commercial Fitbit step-counter). Our app could potentially be used to improve the physical activity among older adults through accurate tracking of their step counts and in-app daily step-count goals.

A catalog of validity indices for step counting wearable technologies during treadmill walking: the CADENCE-Kids study

Background Wearable technologies play an important role in measuring physical activity (PA) and promoting health. Standardized validation indices (i.e., accuracy, bias, and precision) compare performance of step counting wearable technologies in young people. Purpose To produce a catalog of validity indices for step counting wearable technologies assessed during different treadmill speeds (slow [0.8–3.2 km/h], normal [4.0–6.4 km/h], fast [7.2–8.0 km/h]), wear locations (waist, wrist/arm, thigh, and ankle), and age groups (children, 6–12 years; adolescents, 13–17 years; young adults, 18–20 years). Methods One hundred seventeen individuals (13.1 ± 4.2 years, 50.4% female) participated in this cross-sectional study and completed 5-min treadmill bouts (0.8 km/h to 8.0 km/h) while wearing eight devices (Waist: Actical, ActiGraph GT3X+, NL-1000, SW-200; Wrist: ActiGraph GT3X+; Arm: SenseWear; Thigh: activPAL; Ankle: StepWatch). Directly observed steps served as the criterion measure. Accuracy (mean absolute percentage error, MAPE), bias (mean percentage error, MPE), and precision (correlation coefficient, r; standard deviation, SD; coefficient of variation, CoV) were computed. Results Five of the eight tested wearable technologies (i.e., Actical, waist-worn ActiGraph GT3X+, activPAL, StepWatch, and SW-200) performed at < 5% MAPE over the range of normal speeds. More generally, waist (MAPE = 4%), thigh (4%) and ankle (5%) locations displayed higher accuracy than the wrist location (23%) at normal speeds. On average, all wearable technologies displayed the lowest accuracy across slow speeds (MAPE = 50.1 ± 35.5%), and the highest accuracy across normal speeds (MAPE = 15.9 ± 21.7%). Speed and wear location had a significant effect on accuracy and bias (P < 0.001), but not on precision (P > 0.05). Age did not have any effect (P > 0.05). Conclusions Standardized validation indices focused on accuracy, bias, and precision were cataloged by speed, wear location, and age group to serve as important reference points when selecting and/or evaluating device performance in young people moving forward. Reduced performance can be expected at very slow walking speeds (0.8 to 3.2 km/h) for all devices. Ankle-worn and thigh-worn devices demonstrated the highest accuracy. Speed and wear location had a significant effect on accuracy and bias, but not precision. Trial registration Clinicaltrials.govNCT01989104. Registered November 14, 2013.

Abstract 069: Sporadic Step Accumulation And All-cause Mortality In The Women’S Health Study: Do All Steps Count?

Background: With the popularity of step counting and feasibility of accumulating physical activity (PA) through sporadic spurts (e.g., taking the stairs), the 2018 PA Guidelines Committee called for research to inform step-based PA recommendations by quantifying relationships between patterns of stepping and health. Purpose: To examine the relationship between daily steps accumulated outside of “bouts” (sporadic steps/d) and all-cause mortality, before and after accounting for bouted steps/d. Methods: From 2011-2015, 16,732 women (mean 72 [standard deviation 6] years) wore a hip-worn accelerometer for 7 days to assess steps and met wear time criteria. Stepping bouts were defined as ≥10 consecutive minutes at ≥40 steps/min (purposeful stepping or faster), allowing for ≤20% of time and ≤5 mins at <40 steps/min. Total steps/d were partitioned into steps accrued outside of bouts (sporadic steps/d; SS) and in bouts (bouted steps/d; BS). We estimated hazard ratios (HRs) for mortality through Dec 31, 2019 using Cox proportional hazard models fitted to SS in quartiles and using restricted cubic splines. Analyses were adjusted for covariates and repeated with further adjustment for BS, categorized as 0, 1-2000, and >2000 steps/d in bouts. Results: Adjusted HRs (95% confidence intervals) in increasing quartiles of SS were 1.00 (reference); 0.63 (0.52, 0.76); 0.60 (0.49, 0.74); 0.54 (0.42, 0.70). In spline analyses, initial increases in SS corresponded to the greatest mortality reductions (Figure 1), with HRs of 0.69 (0.64, 0.76) per additional 1000 SS below 3200. After further adjusting for BS, initial 1000 steps/d increases in SS were association with HRs of 0.72 (0.66, 0.78). In increasing categories of BS, HRs adjusted for SS were 1.00 (reference); 0.91 (0.76, 1.09); 0.69 (0.56, 0.84). Conclusion: Daily step counts were inversely associated with mortality, regardless of how they were accumulated. These results can help inform step-based target PA volumes that communicate the benefits of increasing everyday walking behaviors.

Step-Counting Function of Adolescent Physical Training APP Based on Artificial Intelligence

With the rapid development of the information age, Internet and other technologies have been making progress, people’s fitness awareness has been gradually enhanced, and sports fitness app has emerged as the times require. This paper mainly studies the step-counting function of physical training app for teenagers based on artificial intelligence. This paper uses the modular development method to achieve the functional requirements of the system as the goal, respectively, for parameter management, website configuration, system log, interface security settings, SMS configuration, WeChat template message and several functional modules to achieve system configuration. In this paper, three types of sensors are used to analyze the data changes in the process of walking through three types of data, and different weights are given as the results of step-counting. When the peak value of sensor data is measured, only the peak value of the primary axial data of each sensor is analyzed, which should be determined according to the actual axial value of the sensor. In this paper, the users’ evaluation indexes of sports fitness app are divided into two groups: importance and satisfaction, so the obtained data are directly divided into two groups: importance and satisfaction of user experience indexes of sports fitness app, and the two groups of data are matched with the sample t test to ensure the scientific conclusion. Finally, the advantages and disadvantages of the user experience of college students’ sports fitness app are analyzed through IPA analysis. Heuristic evaluation is carried out on the step app to score the second-level usability index of the app. The first-level usability index score and the total usability score of the step app are obtained by calculation. There is not much difference between male and female students who use sports apps. Among them, 288 are male students, accounting for 58.2% of the total and 16.4% are female students. The results show that the use of artificial intelligence technology can reduce the overall energy consumption of step-counting algorithm, so as to achieve an energy-saving step-counting algorithm.