Access the full text.
Sign up today, get DeepDyve free for 14 days.
https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 1 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al Background: Gait is an essential manifestation of depression. However, the gait characteristics of daily walking and their relationships with depression have yet to be fully explored. Objective: The aim of this study was to explore associations between depression symptom severity and daily-life gait characteristics derived from acceleration signals in real-world settings. Methods: We used two ambulatory data sets (N=71 and N=215) with acceleration signals collected by wearable devices and mobile phones, respectively. We extracted 12 daily-life gait features to describe the distribution and variance of gait cadence and force over a long-term period. Spearman coefficients and linear mixed-effects models were used to explore the associations between daily-life gait features and depression symptom severity measured by the 15-item Geriatric Depression Scale (GDS-15) and 8-item Patient Health Questionnaire (PHQ-8) self-reported questionnaires. The likelihood-ratio (LR) test was used to test whether daily-life gait features could provide additional information relative to the laboratory gait features. Results: Higher depression symptom severity was significantly associated with lower gait cadence of high-performance walking (segments with faster walking speed) over a long-term period in both data sets. The linear regression model with long-term daily-life gait features (R =0.30) fitted depression scores significantly better (LR test P=.001) than the model with only laboratory gait features (R =0.06). Conclusions: This study indicated that the significant links between daily-life walking characteristics and depression symptom severity could be captured by both wearable devices and mobile phones. The daily-life gait patterns could provide additional information for predicting depression symptom severity relative to laboratory walking. These findings may contribute to developing clinical tools to remotely monitor mental health in real-world settings. (JMIR Mhealth Uhealth 2022;10(10):e40667) doi: 10.2196/40667 KEYWORDS depression; gait; mobile health; mHealth; acceleration signals; monitoring; wearable devices; mobile phones; mental health of wearing sensors on the knees and ankles, for example [14,26]. Introduction Some researchers have suggested that people’s daily-life activity characteristics should have stronger links to their health Depression affects the lives of over 300 million people conditions than laboratory tests [27-29]. Therefore, it is worldwide  and is associated with many adverse outcomes, necessary to monitor and evaluate daily-life walking using including decreased quality of life, loss of occupational function, efficient methods. disability, premature mortality, and suicide [2-5]. While early treatment can be effective and prevent more serious adverse In recent years, several studies have used mobile technologies outcomes , more than half of depressed people do not receive to measure daily-life walking patterns and explored their timely treatment [7,8]. Current questionnaire-based depression associations with depression. However, most of these studies assessments may be affected by recall bias and may not be able only measured the number of cumulative steps of daily-life to collect dynamic information [9,10]. Therefore, several recent walking [30-32], which is more related to individuals’ mobility studies have attempted to explore the associations between and physical activity than to gait patterns (eg, gait cadence and depression and changes in individuals’ behaviors using mobile gait force). To our knowledge, there have been only a few technologies . studies exploring the associations between daily-life gait patterns and depression directly. Adolph et al  found that depressed Changes in gait are essential manifestations of depression participants had reduced walking speed, reduced vertical [12,13]. The main hypothesis linking gait with depression is a up-and-down movements, and more slumped postures compared bidirectional interaction between the brain motor system and with controls by placing two accelerometers on the participant’s cortical and subcortical structures, which are related to emotions trunk and right leg for 2 days . However, wearing multiple and cognitive functions [14-16]. Many studies have explored sensors on the body may not be suitable for long-term the relationships between depression and gait characteristics monitoring. With the development of sensors, the mobile phone based on “gold-standard” laboratory walking tests. Longer gait provides a cost-effective, continuous, and unobtrusive means cycles, reduced stride length, and slower gait cadence were to measure individuals’ behaviors, including daily walking. observed in participants with depression compared with healthy Therefore, the mobile phone may be a potential tool for controls, which have been consistently shown in several studies long-term gait monitoring. [17-25]. Other gait abnormalities such as reduced gait force , increased double support time , reduced swing time The aim of this study was to explore the value of daily-walking variability , slumped postures , and increased body monitoring for improving the evaluation of depression symptom sway  have been reported, but with less consistency across severity. Our first objective was to design and extract gait studies. features from raw acceleration signals to describe the characteristics of daily walking. The second objective was to Laboratory gait tests are hard to be applied in real-world settings explore the associations between gait features and depression because of the need for expensive equipment (eg, video camera symptom severity, and to test whether these associations could and force plates), specialized laboratories, and the inconvenience be captured by different acceleration devices. The third objective https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 2 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al was to test whether daily-life walking could provide additional the United Kingdom (King’s College London [KCL]) between information for predicting depression relative to laboratory November 2017 and April 2021, because the KCL site was the walking. To achieve the second and third objectives, we only site to acquire ethical approval for collecting the phone’s performed our analyses on two ambulatory data sets, the Long acceleration signals. We hereafter denote this subset as the Term Movement Monitoring (LTMM) and Remote Assessment RADAR-MDD-KCL data set for convenience. The phone’s of Disease and Relapse–Major Depressive Disorder acceleration signals were collected at 50 Hz and uploaded to an (RADAR-MDD) data sets [34,35], with acceleration signals open-source platform, RADAR-base . The participants’ collected by a wearable device and mobile phone, respectively. depression symptom severity was assessed by the 8-item Patient Importantly, the LTMM data set contains data related to both Health Questionnaire (PHQ-8)  self-reported through mobile laboratory and daily walking, which could address the third phones every 2 weeks. A patient advisory board comprising study objective. service users co-developed the study. They were involved in the choice of measures, timing, and issues of engagement, and Methods have also been involved in developing the analysis plan. Step Detection Algorithm Data Sets Since we needed to respectively detect steps on the acceleration LTMM Data Set signals collected by wearable devices and mobile phones, we The LTMM data set includes demographics (age and gender), chose to use the step detection algorithm , which was based depression scores (15-item Geriatric Depression Scale [GDS-15] on mobile phones (Figure 1). Given a segment of 3-axis ), and raw acceleration signals (100 Hz) of laboratory acceleration signals (x , y , z ), the magnitude of the acceleration i i i walking tests and 3-day activities for 71 elderly adults , of the segment of acceleration signals was calculated to combine which can be downloaded at PhysioNet . Participants were 3D signals to a single series, r , where . The magnitude included if they did not have any cognitive or gait/balance of the acceleration signals does not depend on the orientation disorders . Participants were asked to walk at a self-selected and tilt of the mobile phone during walking . Subsequently, and comfortable speed for 1 minute in the laboratory while r was filtered by a weighted moving-average filter to remove wearing a 3-axis accelerometer on their lower back . The noise (Equation 1, w=150 milliseconds). Next, the filtered r GDS-15 questionnaire contains 15 easy-to-understand, yes/no format questions, which is suitable for depression screening in was subtracted by the mean of r̄ to make r̄ symmetric to the i i the older population [38,39]. After the laboratory walking test, x-axis. We calculated two new series, B1 and B2 , based on two i i all participants were asked to wear the accelerometer for the thresholds to detect the walking swing phase and stance phase, next 3 consecutive days to record daily activities . respectively (see Equations 2 and 3). If a swing phase ends and a stance phase starts, we can identify a step that occurred. The Ethics Considerations formal detection rule of a step S at sample i is that the following RADAR-MDD was conducted per the Declaration of Helsinki two conditions must be satisfied: (1) a change from –0.5 to 0 and Good Clinical Practice, adhering to principles outlined in in B1 (B1 =0 and B1 =0.5); (2) there is at least one detection i i–1 the National Health Service (NHS) Research Governance of B2=–0.5 in a window of size w=150 milliseconds in sample Framework for Health and Social Care (2nd edition). Ethical i (Min(B2 )=–0.5). i:i+w approval has been obtained in London from the Camberwell St Giles Research Ethics Committee (REC reference 17/LO/1154), in Spain from the CEIC Fundació Sant Joan de Deu (CI PIC-128-17), and in the Netherlands from the Medische Ethische Toetsingscommissie VUms (METc VUmc registratienummer 2018.012–NL63557.029.17). RADAR-MDD Data Set The EU research program RADAR-MDD aimed to investigate the utility of mobile technologies for the long-term monitoring of participants with depression in real-world settings [35,40]. Adult participants with a depression history were included in Then, the gait cycle series could be derived by calculating time the study if they did not meet the following criteria: (1) have intervals between consecutive steps, which was denoted as other psychiatric disorders (eg, bipolar disorder, schizophrenia, Cycles. During each gait cycle, the amplitude from the peak to and dementia), (2) have received treatment for drug or alcohol the valley of the magnitude of the acceleration signals was used use in the past 6 months, (3) a major medical diagnosis that to reflect the gait force of each step. The force of all steps in affects daily activities, and (4) pregnancy . A detailed study the given acceleration signal was denoted as the series Force. protocol was published previously . In this study, we used a subset of RADAR-MDD data collected from a study site in https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 3 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al Figure 1. Step detection algorithm. ACC is the 3-axis acceleration signal; B1 and B2 are two series calculated by thresholds to detect walking swing and stance phase, respectively; and pink dashed lines represent the detected steps. crowded environment or a walking-rest transition status). These Feature Extraction intermittent walking segments may not fully reflect a participant’s normal walking patterns. Therefore, to distinguish Feature Window Size between continuous and intermittent walking, we used a Since the PHQ-8 score is used to estimate depression symptom 1-minute sliding window  to detect steps from the long-term severity for the past 2 weeks , we extracted gait features raw acceleration signals. If the participant was walking most of from a 14-day time window prior to each PHQ-8 record from the time in this minute, we considered this minute as the the RADAR-MDD-KCL data set. For the LTMM data set, we continuous walking segment. Based on our experience, we set extracted gait features from 3-day activities to link daily-life 50 seconds as the threshold for selecting the continuous walking walking with the GDS-15 score. segment; that is, the segment with more than 50 seconds of walking time (sum of all gait cycles in the minute) was selected Step Detection Window and the Continuous Walking for further analysis (Figure 2b). Segment Daily-life walking in real-world settings is complex and contains some intermittent walking segments (such as walking in a https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 4 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al Figure 2. Schematic diagram of long-term gait feature extraction for the Long-Term Movement Monitoring data set. (a) Three-axis acceleration signals of 3 consecutive days; (b) examples of continuous and discontinuous walking segments and three short-term gait features (definitions in Table 1) were extracted from each continuous walking segment; (c) long-term gait feature extraction: 25th percentile, median, 75th percentile, and standard deviation of short-term gait feature values of all continuous walking segments over 3 days for each participant. extracted three short-term gait features from every detected Gait Features continuous walking segment in the feature window. Then, for each short-term gait feature, we calculated four statistical Overview second-order features (long-term features) across all values of The performance of walking varies over time due to several continuous walking segments. In total, 12 long-term gait features factors such as mood, energy, and environment. Therefore, the were extracted in this study, and a summary of these features long-term gait features need to represent the distribution and is shown in Table 1. A schematic diagram of long-term gait variance of walking patterns over the feature window. We first feature extraction is shown in Figure 2. https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 5 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al Table 1. Short-term and long-term gait features extracted and their short descriptions. Gait feature Description Short-term gait features Median cycle (seconds) Median of gait cycles in the 1-minute walking segment Peak frequency (Hz) Peak frequency in the PSD of the magnitude of 1-minute acceleration signals Median of gait force in the 1-minute walking segment Median force (m/s ) Long-term gait features 25th percentile of median cycle 25th percentile of median gait cycle values of all walking segments 50th percentile of median cycle Median of median gait cycle values of all walking segments 75th percentile of median cycle 75th percentile of median gait cycle values of all walking segments SD of median cycle Standard deviation of median gait cycle values of all walking segments 25th percentile of peak frequency 25th percentile of peak frequency values of all walking segments 50th percentile of peak frequency Median of peak frequency values of all walking segments 75th percentile of peak frequency 75th percentile of peak frequency values of all walking segments SD of peak frequency Standard deviation of peak frequency values of all walking segments 25th percentile of median force 25th percentile of median gait force values of all walking segments 50th percentile of median force Median of median gait force values of all walking segments 75th percentile of median force 75th percentile of median gait force values of all walking segments SD of median force Standard deviation of median gait force values of all walking segments PSD: power spectral density (from 0.5 Hz to 3 Hz). All detected continuous walking segments (defined in the Methods section) in a feature window (3 days for the Long Term Movement Monitoring data set and 14 days for the Remote Assessment of Disease and Relapse–Major Depressive Disorder data set). worst walking performance of the participant, which could in Short-Term Gait Features From the 1-Minute Continuous turn reflect physical or mental conditions better than the median Walking Segment value . Therefore, we used 25th percentile, median, and Gait cadence and gait force are essential characteristics of 75th percentile second-order statistics to represent three levels walking. Gait cadence is the rate at which the individual feet of walking performance (low, medium, and high) during a contact the ground . Gait force reflects the ground reaction feature window. For example, faster walking during a feature force during walking . For every continuous walking window could represent high-performance walking, which may segment, the median of the gait cycle series (Cycles) was not be affected by other factors such as fatigue and the crowded calculated to reflect the gait cadence of this minute from the environment. High-performance walking could be represented time domain, which was denoted as median cycle. To assess by the 75th percentile of peak frequency and the 25th percentile the gait cadence from the frequency domain, the power spectral of median cycle in a feature window, which is expected to be density (PSD) of walking was obtained by applying the fast closely associated with depression status. The variance of Fourier transformation to the filtered magnitude (r̄ ) of the daily-life walking in a feature window was measured by the acceleration signals of every continuous walking segment. The SD. peak frequency  of the 0.5-3–Hz band (reflecting walking) Laboratory Gait Features Extracted From Laboratory  of the PSD was used to reflect the main rhythm of steps Walking Tests in the LTMM Data Set from the frequency domain, which was denoted as peak We also extracted median cycle, peak frequency, and median frequency. For gait force, we calculated the median of the Force force from the 1-minute acceleration signals of laboratory series (median force) to represent the average power of all steps walking tests in the LTMM data set. For reading convenience, in the minute. we denoted these as laboratory gait features. Long-Term Gait Features Inclusive Criteria for Data Missingness in the For each of the short-term gait features (median cycle, peak RADAR-MDD-KCL Data Set frequency, and median force), we calculated four statistical second-order features (25th percentile, median, 75th percentile, The raw acceleration signals were remotely collected by mobile and SD) from all detected continuous walking segments during phones in the RADAR-MDD-KCL study. Possibly due to the a feature window. high battery consumption and network traffic for uploading the raw signal, the missing rate of acceleration signals was relatively Previous studies suggested that the extreme values of gait high. To reduce the impact of missingness, a PHQ-8 period (14 characteristics over the long term could reflect the optimal or https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 6 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al days) included in this study should have at least 3 days (aligned included in the RADAR-MDD-KCL data set, the likelihood with the LTMM data set) with more than 50% acceleration ratio test was only performed in the LTMM data set. signals [48,49]. Model A: GDS-15=Age+Gender+3 laboratory gait features (4) Statistical Analyses Model B: GDS-15=Age+Gender+3 laboratory gait For the LTMM data set, Spearman coefficients  were features+12 long-term gait features (5) calculated to assess associations between the GDS-15 score and gait features (3 laboratory gait features and 12 long-term gait Results features). As the data in the RADAR-MDD-KCL data set are longitudinal (repeated PHQ-8 measurements for each Data Summary participant), a series of pairwise linear mixed-effects regression The 71 participants in the LTMM data set have a mean age of models  with random participant intercepts were performed 78.36 (SD 4.71) years with 18 (25%) participants having to explore the association between the PHQ-8 score and each potential depressive disorders (GDS-15≥5) and 69.82 (SD 9.65) of the 12 long-term gait features (no laboratory tests were hours of acceleration signals per participant. The included in the RADAR-MDD-KCL data set). Age, gender, RADAR-MDD-KCL data set, according to the data inclusion and the number of comorbidities (see Table S1 in Multimedia criteria, contains 659 PHQ-8 records collected from 215 Appendix 1) were considered as covariates. The participants and corresponding 99,445 hours (average 463 hours Benjamini-Hochberg method was used for multiple-comparison per participant). The cohort in the RADAR-MDD-KCL data corrections in both data sets . set has a mean age of 43.36 (SD 15.12) years with the majority To test whether long-term gait features could explain additional being women (75%), and half of the PHQ-8 records indicated data variance in depression scores relative to laboratory gait potential depression symptoms (PHQ-8≥10). The average features, we built two nested multivariate linear regression missing rate of acceleration signals collected by phones in the models without and with long-term gait features for the GDS-15 RADAR-MDD-KCL data set (70.60%) was significantly higher score (denoted as Model A and Model B; Equations 4 and 5) than that of the acceleration signals collected by the wearable in the LTMM data set. Specifically, predictors of Model A are device in the LTMM data set (3.03%). A summary of the age, gender, and the 3 laboratory gait features, while predictors demographics, and distributions of depression scores and of Model B are age, gender, the 3 laboratory gait features, and available acceleration signals for participants in the LTMM and the 12 long-term gait features. The coefficient of determination the RADAR-MDD-KCL data sets is shown in Table 2. The (R ) was calculated for both models to estimate how much data heatmaps of correlations between the 12 long-term gait features variance was explained by predictors. Then, the likelihood ratio of the LTMM and RADAR-MDD-KCL data sets are presented test  was used to test whether Model B fit the GDS-15 score in Figure 3. better than Model A. Since the laboratory walking test was not https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 7 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al Table 2. Demographics and distributions of depression scores and available acceleration signals of participants in the two data sets. a b Characteristic LTMM (N=71) RADAR-MDD-KCL (N=215) Age (years), mean (SD) 78.36 (4.71) 43.36 (15.12) Female, n (%) 46 (65%) 162 (75%) c d Depression score, mean (SD) GDS-15 : 3.18 (2.81) PHQ-8 : 9.67 (5.84) 18 (25%) 330 (50%) Potential depressive episode (GDS-15≥5) and PHQ-8≥10), n (%) 71 659 Number of completed depression questionnaires Number of completed depression questionnaires per participant, mean (SD) 1 (0) 3.09 (2.76) Length of total available acceleration signals (hours) 4817 99,445 69.82 (9.65) 98.77 (105.20) Length of available acceleration signals (hours) for each GDS-15/PHQ-8 record , mean (SD) Average missing rate of acceleration signals (%) 3.03 70.60 73.48 (66.98) 113.24 (170.48) Number of continuous walking segments detected from each GDS-15/PHQ-8 record, mean (SD) LTMM: Long Term Movement Monitoring. RADAR-MDD-KCL: subset of the Remote Assessment of Disease and Relapse–Major Depressive Disorder data set collected from King’s College London, United Kingdom. GDS-15: 15-item Geriatric Depression Scale. PHQ-8: 8-item Patient Health Questionnaire. Based on the total number of completed questionnaires. The RADAR-MDD-KCL data set has multiple PHQ-8 records for each participant, which was conducted every 2 weeks. We regarded acceleration signals in the 14 days before a PHQ-8 record. For the GDS-15 record, we considered acceleration signals of all 3-day activities after enrollment. Continuous walking segment was defined as 1-minute acceleration signals with at least 50 seconds of walking (see Methods section). Figure 3. Heatmaps of correlations between 12 long-term gait features of the Long-term Movement Monitoring data set (a) and Remote Assessment of Disease and Relapse–Major Depressive Disorder King's College London data set (b). of gait cycles, lower peak frequency, and smaller median gait Associations Between Gait Features and the GDS-15 force in the 1-minute laboratory walking test. For the long-term Score in the LTMM Data Set period, a higher GDS-15 score was significantly correlated with The Spearman correlations between the GDS-15 score and gait lower variance of gait force and slower cadence of features (both laboratory and long-term gait features) in the high-performance walking and 75th percentile of peak frequency LTMM data set are shown in Table 3. We found that a higher during 3-day activities. GDS-15 score was significantly correlated with a larger median https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 8 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al Table 3. Spearman correlations between the 15-item Geriatric Depression Scale score and gait features, including laboratory and long-term gait features, in the Long-Term Movement Monitoring data set. a b Feature P value Laboratory gait features extracted from the 1-minute laboratory walking test Median cycle 0.39 .001 Peak frequency –0.32 .01 Median force –0.25 .04 Long-term gait feature extracted from 3-day activities 25th percentile of median cycle 0.31 .01 50th percentile of median cycle 0.13 .29 75th percentile of median cycle 0.02 .86 SD of median cycle –0.24 .06 25th percentile of peak frequency –0.02 .85 50th percentile of peak frequency –0.09 .45 75th percentile of peak frequency –0.27 .03 SD of peak frequency –0.12 .33 25th percentile of median force 0.02 .85 50th percentile of median force –0.01 .98 75th percentile of median force –0.10 .41 SD of median force –0.30 .02 Definitions of gait features in this table are provided in Table 1 and the Methods section. P values were adjusted by the Benjamini-Hochberg method for correction of multiple comparisons. high-performance walking, the PHQ-8 score increased by 0.606 Associations Between Long-Term Gait Features and points. Likewise, the 75th percentile of peak frequency was the PHQ-8 Score in the RADAR-MDD-KCL Data Set negatively associated with the PHQ-8 score, indicating that a The pairwise linear mixed-effects models performed in the reduction of 0.1 Hz in the peak frequency of high-performance RADAR-MDD-KCL data set revealed a significant and negative walking was associated with an increase of 0.26 PHQ-8 points. link between the PHQ-8 score and the gait cadence of Other long-term gait features were not found to be significantly high-performance walking during the 14 days before submitting associated with the PHQ-8 score in the RADAR-MDD-KCL PHQ-8 records. Specifically, the 25th percentile of median cycle data set. A summary of all 12 linear mixed-effects regression was positively associated with the PHQ-8 score; that is, for models is provided in Table 4. every increase of 0.1 seconds in the median gait cycle of https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 9 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al Table 4. Twelve pairwise linear mixed-effects models for exploring associations between long-term gait features and depression symptom severity (8-item Patient Health Questionnaire) in the RADAR-MDD-KCL data set. b c Estimate SE df t value Long-term gait feature P value 25th percentile of median cycle 6.06 2.72 648.75 2.23 .03 50th percentile of median cycle 3.98 2.51 639.41 1.59 .11 75th percentile of median cycle 2.49 2.08 653.72 1.20 .23 STD of median cycle 2.87 4.41 631.11 0.65 .52 25th percentile of peak frequency –1.50 1.02 656.44 –1.46 .15 50th percentile of peak frequency –1.93 1.05 650.76 –1.83 .07 75th percentile of peak frequency –2.62 1.01 634.70 –2.60 .01 SD of peak frequency 0.21 1.86 600.50 0.12 .91 25th percentile of median force –0.57 2.24 637.46 –0.25 .80 50th percentile of median force 0.88 1.79 655.77 0.49 .62 75th percentile of median force 0.44 1.66 656.37 0.26 .79 SD of median force 2.05 3.78 602.90 0.54 .59 RADAR-MDD-KCL: Subset of Remote Assessment of Disease and Relapse–Major Depressive Disorder collected from King’s College London. Definitions of daily-life gait features are provided in Table 1 and the Methods section. P values were adjusted by the Benjamini-Hochberg method for correction of multiple comparisons. The results of Spearman correlations between laboratory gait Results of the Likelihood Ratio Test in the LTMM features and the GDS-15 score in the LTMM data set are Data Set consistent with previous studies [17-25]; that is, the participants The regression model with long-term gait features (Model B) with more severe depression symptoms were more likely to have slower gait cadence (longer median of gait cycles and achieved better performance (R =0.30) than the model without 2 lower gait frequency) and smaller gait force in laboratory long-term gait features (Model A) (R =0.06). We found that the walking tests. 12 long-term gait features extracted from 3-day activities could For daily-life walking, this study used the faster walking (75th explain an extra 24% data variance (an increase of 0.24 in R ) percentile of peak frequency and 25th percentile of median of GDS-15 scores relative to the laboratory gait features and cycle) in all detected continuous walking segments to represent participants’ demographics. The likelihood ratio test showed high-performance walking during a feature window (3 days for that Model B fitted GDS-15 scores significantly better than 2 2 LTMM and 14 days for RADAR-MDD-KCL). Only gait Model A (χ =32.91>χ (12), P=.001). The detailed results 0.05 cadence of high-performance walking was found to be of the two nested regression models are shown in Table S2 of significantly and negatively associated with depression symptom Multimedia Appendix 1. severity, whereas gait patterns under medium/low-performance walking were not significantly associated with the depression Discussion score. This finding was consistent in both the LTMM and RADAR-MDD-KCL data sets. A potential reason is that the Principal Findings walking performance in real-world scenarios may be affected This study retrospectively used two ambulatory data sets for by multiple factors (such as walking during the day or at night, exploring the associations between depression symptom severity walking under fatigue or walking after rest, and walking to a and daily-life gait characteristics. We extracted 12 long-term destination or navigating a crowded supermarket) ; gait features to describe the distribution and variance of gait therefore, the lower walking performance may not fully reflect cadence and force over a long-term period and link daily-life the participant’s physical or mental conditions. Therefore, from gait patterns with a self-reported depression score. The main the main finding of this study, we speculated that faster steps findings of this study are (1) higher depression symptom severity over a long-term period could represent the high performance is significantly associated with lower gait cadence of of participants’ walking, which could be closely associated with high-performance walking (faster walking in all continuous their depression status. walking segments) over a long-term period; (2) long-term In the LTMM data set, we found that the variance of gait force daily-life walking has the potential to provide additional (SD of median force) in 3-day activities was significantly and information for predicting depression symptom severity relative negatively associated with the depression symptom severity, to laboratory gait characteristics and demographics; and (3) indicating that participants with higher depression symptom wearable devices and mobile phones both have potential to severity were likely to have relatively monotonous walking over capture the associations between daily gait and depression. 3 days. However, the feature was not significantly associated https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 10 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al with the PHQ-8 score in the RADAR-MDD-KCL data set. One raw acceleration signals, and the Android operating system reason is that the magnitude (r ) (explained in the Step Detection moderation of resources. According to the findings of this study, a possible solution to reduce missingness is uploading gait cycles Algorithm section) of the acceleration signals depends on the instead of uploading raw acceleration signals in future long-term location of the accelerometers attached to the body . As monitoring research. This is not difficult to implement, as most acceleration signals in the RADAR-MDD-KCL data set were current smartphones have real-time step detection functions or collected by mobile phones, the variable locations of phones apps [56,57]. Furthermore, the self-reported PHQ-8 data may when attached to participants’ bodies (such as in the hand, be subject to recall bias. We may consider implementing handbag, and pocket) affected the magnitude of acceleration ecological momentary assessments with passive gait data signals. Therefore, the magnitude of phone-collected collection in future research. acceleration signals cannot fully reflect the gait force. The hyperparameters in step detection and feature extraction Results of regression models and the likelihood test in the need further investigation. We considered using a 1-minute LTMM data set illustrated the importance of monitoring window size for step detection and 50 seconds for continuous daily-life gait in real-world settings. Laboratory gait features walking segment selection based on previous studies [34,44] and demographics in LTMM data only explained a small and our experience. The feature window sizes for the two data proportion of data variance of the GDS-15 score (R =0.06), sets are different due to the different study designs. However, whereas long-term gait features extracted from 3-day activities the optimal hyperparameters are still unclear and will be could explain an extra 24% of data variance (R =0.30). This investigated in future research. finding supported that long-term daily-life walking has the potential to provide additional information for predicting Gait features extracted in this study were simple and statistically depression symptom severity relative to laboratory gait based, which were used to illustrate the importance of daily characteristics and demographics. Further, this finding also walking in our initial analysis. More features such as nonlinear indicated that the laboratory walking test may be affected by features will be considered in future research. several factors such as subjective psychological factors and Gait characteristics could be affected by some physical diseases, laboratory-controlled conditions, which may not fully reflect neurological disorders, and age [58-60]. Although none of the the condition of a participant’s mental health [27,29]. Since participants had any cognitive or gait/balance disorders in the there were no laboratory tests in the RADAR-MDD-KCL data LTMM data set and the number of comorbidities and age were set, the comparison between laboratory gait features and considered as covariates in the RADAR-MDD-KCL data set, long-term daily-life gait features was not performed in the physical comorbidities and other comorbidities may have RADAR-MDD-KCL. We will consider adding laboratory tests different impacts on the gait characteristics. We will consider at enrollment in future digital depression studies. a wider range of comorbidities and investigate them further in future research. Limitations Although we found that wearables and mobile phones have the Conclusion potential to capture the associations between depression and In summary, the findings of this study showed that significant daily-life gait patterns, both devices have some limitations. links between depression symptom severity and daily-life gait Wearables could collect relatively complete walking data; characteristics could be captured in different data sets and by however, wearing sensors may not be suitable for long-term different accelerometer devices. Long-term daily-life walking monitoring. Mobile phones could be used for long-term patterns could provide additional value for understanding monitoring without user burden, but the missing rate of mobile depression manifestations relative to gait patterns in laboratory phone acceleration signals is relatively high. The findings of walking tests, which illustrated the importance of long-term this study support that the links between gait and depression gait monitoring. The gait cadence of high-performance walking could still be revealed from the limited and sparse daily-life in daily life has the potential to be an indicator for monitoring walking acceleration signals. Missingness is a common depression severity, which may contribute to developing clinical challenge in remote digital studies , which may be caused tools to remotely monitor mental health in real-world settings. by high battery consumption, network traffic for uploading the Acknowledgments The Remote Assessment of Disease and Relapse–Central Nervous System (RADAR-CNS) project has received funding from the Innovative Medicines Initiative (IMI) 2 Joint Undertaking under grant agreement number 115902. This Joint Undertaking receives support from the European Union’s Horizon 2020 Research and Innovation Program and the European Federation of Pharmaceutical Industries and Associations (EFPIA). This communication reflects the views of the RADAR-CNS consortium and neither IMI nor the European Union and EFPIA are liable for any use that may be made of the information contained herein. The funding bodies have not been involved in the design of the study, the collection or analysis of data, or the interpretation of data. This study represents independent research partly funded by the National Institute for Health Research (NIHR) Maudsley Biomedical Research Centre at South London, and Maudsley NHS Foundation Trust and King’s College London. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care. We thank all the members of the RADAR-CNS patient advisory board for their contribution to the device selection https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 11 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al procedures, and their invaluable advice throughout the study protocol design. This research was reviewed by a team with experience of mental health problems and their careers, who have been specially trained to advise on research proposals and documentation through Feasibility and Acceptability Support Team for Researchers (FAST-R), a free, confidential service in England provided by the NIHR Maudsley Biomedical Research Centre via King’s College London and South London and Maudsley NHS Foundation Trust. We thank all GLAD Study volunteers for their participation, and gratefully acknowledge the NIHR BioResource, NIHR BioResource centers, NHS Trusts and staff for their contribution. We also acknowledge NIHR BRC, King’s College London, South London and Maudsley NHS Trust and King’s Health Partners. We thank the NIHR, NHS Blood and Transplant, and Health Data Research UK as part of the Digital Innovation Hub Program. CO is supported by the UK Medical Research Council (MR/N013700/1) and King’s College London member of the MRC Doctoral Training Partnership in Biomedical Sciences. RJBD is supported by the following: (1) NIHR Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London, United Kingdom; (2) Health Data Research UK, which is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care (England), Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division (Welsh Government), Public Health Agency (Northern Ireland), British Heart Foundation, and Wellcome Trust; (3) The BigData@Heart consortium, funded by the IMI-2 Joint Undertaking under grant agreement number 116074. This Joint Undertaking receives support from the European Union’s Horizon 2020 Research and Innovation Program and EFPIA; it is chaired by DE Grobbee and SD Anker, partnering with 20 academic and industry partners and ESC; (4) the NIHR University College London Hospitals Biomedical Research Centre; (5) the NIHR Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London; (6) the UK Research and Innovation London Medical Imaging and Artificial Intelligence Centre for Value Based Healthcare; and (7) the NIHR Applied Research Collaboration South London (NIHR ARC South London) at King’s College Hospital NHS Foundation Trust. Conflicts of Interest SV and VAN are employees of Janssen Research and Development LLC. PA is employed by the pharmaceutical company H Lundbeck A/S. DCM has accepted honoraria and consulting fees from Otsuka Pharmaceuticals, Optum Behavioral Health, Centerstone Research Institute, and the One Mind Foundation; has received royalties from Oxford Press; and has an ownership interest in Adaptive Health, Inc. MH is the principal investigator of RADAR-CNS, a private public precompetitive consortium that receives funding from Janssen, UCB, Lundbeck, MSD, and Biogen. The other authors have no conflicts of interest to declare. Multimedia Appendix 1 Tables S1-S2. [DOCX File , 22 KB-Multimedia Appendix 1] References 1. Friedrich M. Depression Is the Leading Cause of Disability Around the World. JAMA 2017 Apr 18;317(15):1517. [doi: 10.1001/jama.2017.3826] [Medline: 28418490] 2. Lenox-Smith A, Macdonald MTB, Reed C, Tylee A, Peveler R, Quail D, et al. Quality of life in depressed patients in UK primary care: The FINDER Study. Neurol Ther 2013 Dec 13;2(1-2):25-42 [FREE Full text] [doi: 10.1007/s40120-013-0006-1] [Medline: 26000214] 3. Lerner D, Adler D, Chang H, Berndt ER, Irish JT, Lapitsky L, et al. The clinical and occupational correlates of work productivity loss among employed patients with depression. J Occup Environ Med 2004 Jun;46(6 Suppl):S46-S55 [FREE Full text] [doi: 10.1097/01.jom.0000126684.82825.0a] [Medline: 15194895] 4. Lewinsohn PM, Solomon A, Seeley JR, Zeiss A. Clinical implications of "subthreshold" depressive symptoms. J Abnorm Psychol 2000 May;109(2):345-351. [Medline: 10895574] 5. Hawton K, Casañas I Comabella C, Haw C, Saunders K. Risk factors for suicide in individuals with depression: a systematic review. J Affect Disord 2013 May;147(1-3):17-28. [doi: 10.1016/j.jad.2013.01.004] [Medline: 23411024] 6. Kamphuis MH, Stegenga BT, Zuithoff NPA, King M, Nazareth I, de Wit NJ, et al. Does recognition of depression in primary care affect outcome? The PREDICT-NL study. Fam Pract 2012 Feb 22;29(1):16-23. [doi: 10.1093/fampra/cmr049] [Medline: 21859837] 7. Harman JS, Edlund MJ, Fortney JC. Disparities in the adequacy of depression treatment in the United States. Psychiatr Serv 2004 Dec;55(12):1379-1385. [doi: 10.1176/appi.ps.55.12.1379] [Medline: 15572565] 8. Young AS, Klap R, Sherbourne CD, Wells KB. The quality of care for depressive and anxiety disorders in the United States. Arch Gen Psychiatry 2001 Jan 01;58(1):55-61. [doi: 10.1001/archpsyc.58.1.55] [Medline: 11146758] 9. Althubaiti A. Information bias in health research: definition, pitfalls, and adjustment methods. J Multidiscip Healthc 2016 May;9:211-217. [doi: 10.2147/JMDH.S104807] [Medline: 27217764] https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 12 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al 10. Devaux M, Sassi F. Social disparities in hazardous alcohol use: self-report bias may lead to incorrect estimates. Eur J Public Health 2016 Feb 19;26(1):129-134 [FREE Full text] [doi: 10.1093/eurpub/ckv190] [Medline: 26585784] 11. Rohani DA, Faurholt-Jepsen M, Kessing LV, Bardram JE. Correlations between objective behavioral features collected from mobile and wearable devices and depressive mood symptoms in patients with affective disorders: systematic review. JMIR Mhealth Uhealth 2018 Aug 13;6(8):e165 [FREE Full text] [doi: 10.2196/mhealth.9691] [Medline: 30104184] 12. Sobin C, Sackeim HA. Psychomotor symptoms of depression. Am J Psychiatry 1997 Jan;154(1):4-17. [doi: 10.1176/ajp.154.1.4] [Medline: 8988952] 13. Schrijvers D, Hulstijn W, Sabbe BG. Psychomotor symptoms in depression: a diagnostic, pathophysiological and therapeutic tool. J Affect Disord 2008 Jul;109(1-2):1-20. [doi: 10.1016/j.jad.2007.10.019] [Medline: 18082896] 14. Deligianni F, Guo Y, Yang G. From emotions to mood disorders: a survey on gait analysis methodology. IEEE J Biomed Health Inform 2019 Nov;23(6):2302-2316. [doi: 10.1109/jbhi.2019.2938111] 15. Sanders RD, Gillig PM. Gait and its assessment in psychiatry. Psychiatry 2010 Jul;7(7):38-43 [FREE Full text] [Medline: 20805918] 16. Yogev-Seligmann G, Hausdorff JM, Giladi N. The role of executive function and attention in gait. Mov Disord 2008 Feb 15;23(3):329-42; quiz 472 [FREE Full text] [doi: 10.1002/mds.21720] [Medline: 18058946] 17. Lemke MR, Wendorff T, Mieth B, Buhl K, Linnemann M. Spatiotemporal gait patterns during over ground locomotion in major depression compared with healthy controls. J Psychiatr Res 2000;34(4-5):277-283. [doi: 10.1016/s0022-3956(00)00017-0] [Medline: 11104839] 18. Michalak J, Troje N, Heidenreich T. The effects of mindfulness-based cognitive therapy on depressive gait patterns. J Cogn Behav Psychother 2011;11(1):13-27. 19. van Iersel MB, Haitsma A, Olde Rikkert MGM, Benraad C. Quantitative gait analysis to detect gait disorders in geriatric patients with depression. J Am Geriatr Soc 2005 Aug;53(8):1441-1442. [doi: 10.1111/j.1532-5415.2005.53433_4.x] [Medline: 16078982] 20. Sanders JB, Bremmer MA, Comijs HC, Deeg DJ, Beekman AT. Gait speed and the natural course of depressive symptoms in late life; an independent association with chronicity? J Am Med Dir Assoc 2016 Apr 01;17(4):331-335. [doi: 10.1016/j.jamda.2015.11.016] [Medline: 26775581] 21. Sloman L, Pierrynowski M, Berridge M, Tupling S, Flowers J. Mood, depressive illness and gait patterns. Can J Psychiatry 1987 Apr 01;32(3):190-193. [doi: 10.1177/070674378703200306] [Medline: 3567834] 22. Radovanović S, Jovičić M, Marić NP, Kostić V. Gait characteristics in patients with major depression performing cognitive and motor tasks while walking. Psychiatry Res 2014 Jun 30;217(1-2):39-46. [doi: 10.1016/j.psychres.2014.02.001] [Medline: 24613201] 23. Brandler TC, Wang C, Oh-Park M, Holtzer R, Verghese J. Depressive symptoms and gait dysfunction in the elderly. Am J Geriatr Psychiatry 2012 May;20(5):425-432 [FREE Full text] [doi: 10.1097/JGP.0b013e31821181c6] [Medline: 21422907] 24. Michalak J, Troje N, Fischer J, Vollmar P, Heidenreich T, Schulte D. Embodiment of sadness and depression--gait patterns associated with dysphoric mood. Psychosom Med 2009 Jun;71(5):580-587. [doi: 10.1097/PSY.0b013e3181a2515c] [Medline: 19414617] 25. Pieruccini-Faria F, Muir-Hunter SW, Montero-Odasso M. Do depressive symptoms affect balance in older adults with mild cognitive impairment? Results from the "gait and brain study". Exp Gerontol 2018 Jul 15;108:106-111. [doi: 10.1016/j.exger.2018.04.004] [Medline: 29653157] 26. Wang Y, Wang J, Liu X, Zhu T. Detecting depression through gait data: examining the contribution of gait features in recognizing depression. Front Psychiatry 2021 May 7;12:661213. [doi: 10.3389/fpsyt.2021.661213] [Medline: 34025483] 27. Notthoff N, Drewelies J, Kazanecka P, Steinhagen-Thiessen E, Norman K, Düzel S, et al. Feeling older, walking slower-but only if someone's watching. Subjective age is associated with walking speed in the laboratory, but not in real life. Eur J Ageing 2018 Dec 1;15(4):425-433 [FREE Full text] [doi: 10.1007/s10433-017-0450-3] [Medline: 30532679] 28. Atrsaei A, Corrà MF, Dadashi F, Vila-Chã N, Maia L, Mariani B, et al. Gait speed in clinical and daily living assessments in Parkinson's disease patients: performance versus capacity. NPJ Parkinsons Dis 2021 Mar 05;7(1):24. [doi: 10.1038/s41531-021-00171-0] [Medline: 33674597] 29. Rispens SM, van Schooten KS, Pijnappels M, Daffertshofer A, Beek PJ, van Dieën JH. Do extreme values of daily-life gait characteristics provide more information about fall risk than median values? JMIR Res Protoc 2015 Jan 05;4(1):e4 [FREE Full text] [doi: 10.2196/resprot.3931] [Medline: 25560937] 30. McKercher CM, Schmidt MD, Sanderson KA, Patton GC, Dwyer T, Venn AJ. Physical activity and depression in young adults. Am J Prev Med 2009 Feb;36(2):161-164. [doi: 10.1016/j.amepre.2008.09.036] [Medline: 19062235] 31. Abedi P, Nikkhah P, Najar S. Effect of pedometer-based walking on depression, anxiety and insomnia among postmenopausal women. Climacteric 2015 Sep 23;18(6):841-845. [doi: 10.3109/13697137.2015.1065246] [Medline: 26100101] 32. Große J, Petzold MB, Brand R, Ströhle A. Step away from depression-Study protocol for a multicenter randomized clinical trial for a pedometer intervention during and after in-patient treatment of depression. Int J Methods Psychiatr Res 2021 Mar 12;30(1):e1862 [FREE Full text] [doi: 10.1002/mpr.1862] [Medline: 33180994] 33. Adolph D, Tschacher W, Niemeyer H, Michalak J. Gait patterns and mood in everyday life: a comparison between depressed patients and non-depressed controls. Cogn Ther Res 2021 Feb 25;45(6):1128-1140. [doi: 10.1007/s10608-021-10215-7] https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 13 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al 34. Weiss A, Brozgol M, Dorfman M, Herman T, Shema S, Giladi N, et al. Does the evaluation of gait quality during daily life provide insight into fall risk? A novel approach using 3-day accelerometer recordings. Neurorehabil Neural Repair 2013 Oct 17;27(8):742-752. [doi: 10.1177/1545968313491004] [Medline: 23774124] 35. Matcham F, Barattieri di San Pietro C, Bulgari V, de Girolamo G, Dobson R, Eriksson H, RADAR-CNS consortium. Remote assessment of disease and relapse in major depressive disorder (RADAR-MDD): a multi-centre prospective cohort study protocol. BMC Psychiatry 2019 Feb 18;19(1):72 [FREE Full text] [doi: 10.1186/s12888-019-2049-z] [Medline: 30777041] 36. D'Ath P, Katona P, Mullan E, Evans S, Katona C. Screening, detection and management of depression in elderly primary care attenders. I: The acceptability and performance of the 15 item Geriatric Depression Scale (GDS15) and the development of short versions. Fam Pract 1994 Sep;11(3):260-266. [doi: 10.1093/fampra/11.3.260] [Medline: 7843514] 37. Goldberger AL, Amaral LAN, Glass L, Hausdorff JM, Ivanov PC, Mark RG, et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 2000 Jun 13;101(23):E215-E220. [doi: 10.1161/01.cir.101.23.e215] [Medline: 10851218] 38. Yesavage JA, Sheikh JI. 9/Geriatric Depression Scale (GDS). Clinical Gerontologist 2008 Oct 25;5(1-2):165-173. [doi: 10.1300/J018v05n01_09] 39. Williams EI, Wallace P. Health checks for people aged 75 and over. Occas Pap R Coll Gen Pract 1993 Apr(59):1-30 [FREE Full text] [Medline: 8199581] 40. Matcham F, Leightley D, Siddi S, Lamers F, White KM, Annas P, RADAR-CNS consortium. Remote Assessment of Disease and Relapse in Major Depressive Disorder (RADAR-MDD): recruitment, retention, and data availability in a longitudinal remote measurement study. BMC Psychiatry 2022 Feb 21;22(1):136 [FREE Full text] [doi: 10.1186/s12888-022-03753-1] [Medline: 35189842] 41. Ranjan Y, Rashid Z, Stewart C, Conde P, Begale M, Verbeeck D, Hyve, RADAR-CNS Consortium. RADAR-Base: open source mobile health platform for collecting, monitoring, and analyzing data using sensors, wearables, and mobile devices. JMIR Mhealth Uhealth 2019 Aug 01;7(8):e11734 [FREE Full text] [doi: 10.2196/11734] [Medline: 31373275] 42. Kroenke K, Strine TW, Spitzer RL, Williams JBW, Berry JT, Mokdad AH. The PHQ-8 as a measure of current depression in the general population. J Affect Disord 2009 Apr;114(1-3):163-173. [doi: 10.1016/j.jad.2008.06.026] [Medline: 18752852] 43. Marron J, Labrador M, Valle A, Lanvin D, Rodriguez M. Multi sensor system for pedestrian tracking and activity recognition in indoor environments. IJAHUC 2016;23(1/2):3-23. [doi: 10.1504/IJAHUC.2016.078480] 44. Ihlen EAF, Weiss A, Helbostad JL, Hausdorff JM. The discriminant value of phase-dependent local dynamic stability of daily life walking in older adult community-dwelling fallers and nonfallers. Biomed Res Int 2015;2015:402596-402511. [doi: 10.1155/2015/402596] [Medline: 26491669] 45. Whittle M. Gait analysis. In: McLatchie GR, Lennox CME, editors. The soft tissues. Amsterdam: Elsevier; 1993:187-199. 46. Herzog W, Nigg BM, Read LJ, Olsson E. Asymmetries in ground reaction force patterns in normal human gait. Med Sci Sports Exerc 1989 Feb;21(1):110-114. [doi: 10.1249/00005768-198902000-00020] [Medline: 2927295] 47. Sun L, Zheng X, Shou H, Li J, Li Y. Quantitative prediction of channel sand bodies based on seismic peak attributes in the frequency domain and its application. Appl. Geophys 2010 Jun 10;7(1):10-17. [doi: 10.1007/s11770-010-0009-y] 48. Zhang Y, Folarin AA, Sun S, Cummins N, Ranjan Y, Rashid Z, et al. Predicting depressive symptom severity through individuals' nearby bluetooth device count data collected by mobile phones: preliminary longitudinal study. JMIR Mhealth Uhealth 2021 Jul 30;9(7):e29840 [FREE Full text] [doi: 10.2196/29840] [Medline: 34328441] 49. Saeb S, Zhang M, Kwasny MM, Karr CJ, Kording K, Mohr DC. The relationship between clinical, momentary, and sensor-based assessment of depression. Int Conf Pervasive Comput Technol Healthc 2015 Aug;2015:229-232 [FREE Full text] [doi: 10.4108/icst.pervasivehealth.2015.259034] [Medline: 26640739] 50. Spearman C. The proof and measurement of association between two things. Am J Psychol 1987;100(3/4):441. [doi: 10.2307/1422689] 51. Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics 1982 Dec;38(4):963. [doi: 10.2307/2529876] 52. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B 2018 Dec 05;57(1):289-300. [doi: 10.1111/j.2517-6161.1995.tb02031.x] 53. Glover S, Dixon P. Likelihood ratios: a simple and flexible statistic for empirical psychologists. Psychon Bull Rev 2004 Oct;11(5):791-806. [doi: 10.3758/bf03196706] [Medline: 15732688] 54. Derawi M, Bours P. Gait and activity recognition using commercial phones. Comput Secur 2013 Nov;39:137-144. [doi: 10.1016/j.cose.2013.07.004] 55. Onnela J. Opportunities and challenges in the collection and analysis of digital phenotyping data. Neuropsychopharmacology 2021 Jan 17;46(1):45-54 [FREE Full text] [doi: 10.1038/s41386-020-0771-3] [Medline: 32679583] 56. Silva AG, Simões P, Queirós A, Rodrigues M, Rocha NP. Mobile apps to quantify aspects of physical activity: a systematic review on its reliability and validity. J Med Syst 2020 Jan 08;44(2):51. [doi: 10.1007/s10916-019-1506-z] [Medline: 31915935] 57. Stavropoulos T, Andreadis S, Mpaltadoros L, Nikolopoulos S, Kompatsiaris I. Wearable sensors and smartphone apps as pedometers in eHealth: a comparative accuracy, reliability and user evaluation. 2020 Presented at: 2020 IEEE International Conference on Human-Machine Systems (ICHMS); September 7-9, 2020; online. [doi: 10.1109/ichms49158.2020.9209441] https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 14 (page number not for citation purposes) XSL FO RenderX JMIR MHEALTH AND UHEALTH Zhang et al 58. Del Din S, Godfrey A, Galna B, Lord S, Rochester L. Free-living gait characteristics in ageing and Parkinson's disease: impact of environment and ambulatory bout length. J Neuroeng Rehabil 2016 May 12;13(1):46 [FREE Full text] [doi: 10.1186/s12984-016-0154-5] [Medline: 27175731] 59. Rodgers MM, Mulcare JA, King DL, Mathews T, Gupta SC, Glaser RM. Gait characteristics of individuals with multiple sclerosis before and after a 6-month aerobic training program. J Rehabil Res Dev 1999 Jul;36(3):183-188. [Medline: 10659801] 60. Helbostad JL, Leirfall S, Moe-Nilssen R, Sletvold O. Physical fatigue affects gait characteristics in older persons. J Gerontol A Biol Sci Med Sci 2007 Sep 01;62(9):1010-1015. [doi: 10.1093/gerona/62.9.1010] [Medline: 17895440] Abbreviations EFPIA: European Federation of Pharmaceutical Industries and Associations FAST-R: Feasibility and Acceptability Support Team for Researchers GDS-15: 15-item Geriatric Depression Scale IMI: Innovative Medicines Initiative LR: likelihood ratio KCL: King’s College London LTMM: Long Term Movement Monitoring NHS: National Health Service NIHR: National Institute of Health Research PHQ-8: 8-item Patient Health Questionnaire PSD: power spectral density RADAR-CNS: Remote Assessment of Disease and Relapse–Central Nervous System RADAR-MDD: Remote Assessment of Disease and Relapse–Major Depressive Disorder Edited by L Buis; submitted 30.06.22; peer-reviewed by Z Yang, D Lekkas; comments to author 02.08.22; revised version received 11.08.22; accepted 26.08.22; published 04.10.22 Please cite as: Zhang Y, Folarin AA, Sun S, Cummins N, Vairavan S, Qian L, Ranjan Y, Rashid Z, Conde P, Stewart C, Laiou P, Sankesara H, Matcham F, White KM, Oetzmann C, Ivan A, Lamers F, Siddi S, Simblett S, Rintala A, Mohr DC, Myin-Germeys I, Wykes T, Haro JM, Penninx BWJH, Narayan VA, Annas P, Hotopf M, Dobson RJB, RADAR-CNS Consortium Associations Between Depression Symptom Severity and Daily-Life Gait Characteristics Derived From Long-Term Acceleration Signals in Real-World Settings: Retrospective Analysis JMIR Mhealth Uhealth 2022;10(10):e40667 URL: https://mhealth.jmir.org/2022/10/e40667 doi: 10.2196/40667 PMID: ©Yuezhou Zhang, Amos A Folarin, Shaoxiong Sun, Nicholas Cummins, Srinivasan Vairavan, Linglong Qian, Yatharth Ranjan, Zulqarnain Rashid, Pauline Conde, Callum Stewart, Petroula Laiou, Heet Sankesara, Faith Matcham, Katie M White, Carolin Oetzmann, Alina Ivan, Femke Lamers, Sara Siddi, Sara Simblett, Aki Rintala, David C Mohr, Inez Myin-Germeys, Til Wykes, Josep Maria Haro, Brenda W J H Penninx, Vaibhav A Narayan, Peter Annas, Matthew Hotopf, Richard J B Dobson, RADAR-CNS Consortium. Originally published in JMIR mHealth and uHealth (https://mhealth.jmir.org), 04.10.2022. This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mHealth and uHealth, is properly cited. The complete bibliographic information, a link to the original publication on https://mhealth.jmir.org/, as well as this copyright and license information must be included. https://mhealth.jmir.org/2022/10/e40667 JMIR Mhealth Uhealth 2022 | vol. 10 | iss. 10 | e40667 | p. 15 (page number not for citation purposes) XSL FO RenderX
JMIR mHealth and uHealth – JMIR Publications
Published: Oct 4, 2022
Keywords: depression; gait; mobile health; mHealth; acceleration signals; monitoring; wearable devices; mobile phones; mental health
Access the full text.
Sign up today, get DeepDyve free for 14 days.