This paper is in the following e-collection/theme issue:

Digital Biomarkers and Digital Phenotyping (215) Wearable Devices and Sensors (947) mHealth for Data Collection and Research (1035) Digital Mental Health Interventions, e-Mental Health and Cyberpsychology (1701) Psychotic Disorders (172) Mobile Health in Psychiatry (463)

Published on in Vol 13 (2025)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/56185, first published .
Utility of Digital Phenotyping Based on Wrist Wearables and Smartphones in Psychosis: Observational Study

Utility of Digital Phenotyping Based on Wrist Wearables and Smartphones in Psychosis: Observational Study

Utility of Digital Phenotyping Based on Wrist Wearables and Smartphones in Psychosis: Observational Study

Authors of this article:

Zixu Yang1 Author Orcid Image ;   Creighton Heaukulani2 Author Orcid Image ;   Amelia Sim3 Author Orcid Image ;   Thisum Buddhika2 Author Orcid Image ;   Nur Amirah Abdul Rashid1 Author Orcid Image ;   Xuancong Wang2 Author Orcid Image ;   Shushan Zheng3 Author Orcid Image ;   Yue Feng Quek1 Author Orcid Image ;   Sutapa Basu3 Author Orcid Image ;   Kok Wei Lee3 Author Orcid Image ;   Charmaine Tang3 Author Orcid Image ;   Swapna Verma3 Author Orcid Image ;   Robert J T Morris2, 4 Author Orcid Image ;   Jimmy Lee1, 3, 5 Author Orcid Image
Article Authors Cited by Tweetations (5) Metrics

    1North Region, Institute of Mental Health, Singapore, Singapore

    2Ministry of Health Office for Healthcare Transformation, Singapore, Singapore

    3Department of Psychosis, Institute of Mental Health, Singapore, Singapore

    4Yong Loo Lin School of Medicine, National University of Singapore, Singapore, Singapore

    5Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Singapore

    *these authors contributed equally

    Corresponding Author:

    Jimmy Lee, MBBS, MMed, MCI


    Background: Digital phenotyping provides insights into an individual’s digital behaviors and has potential clinical utility.

    Objective: In this observational study, we explored digital biomarkers collected from wrist-wearable devices and smartphones and their associations with clinical symptoms and functioning in patients with schizophrenia.

    Methods: We recruited 100 outpatients with schizophrenia spectrum disorder, and we collected various digital data from commercially available wrist wearables and smartphones over a 6-month period. In this report, we analyzed the first week of digital data on heart rate, sleep, and physical activity from the wrist wearables and travel distance, sociability, touchscreen tapping speed, and screen time from the smartphones. We analyzed the relationships between these digital measures and patient baseline measurements of clinical symptoms assessed with the Positive and Negative Syndrome Scale, Brief Negative Symptoms Scale, and Calgary Depression Scale for Schizophrenia, as well as functioning as assessed with the Social and Occupational Functioning Assessment Scale. Linear regression was performed for each digital and clinical measure independently, with the digital measures being treated as predictors.

    Results: Digital data were successfully collected from both the wearables and smartphones throughout the study, with 91% of the total possible data successfully collected from the wearables and 82% from the smartphones during the first week of the trial—the period under analysis in this report. Among the clinical outcomes, negative symptoms were associated with the greatest number of digital measures (10 of the 12 studied here), followed by overall measures of psychopathology symptoms, functioning, and positive symptoms, which were each associated with at least 3 digital measures. Cognition and cognitive/disorganization symptoms were each associated with 1 or 2 digital measures.

    Conclusions: We found significant associations between nearly all digital measures and a wide range of symptoms and functioning in a community sample of individuals with schizophrenia. These findings provide insights into the digital behaviors of individuals with schizophrenia and highlight the potential of using commercially available wrist wearables and smartphones for passive monitoring in schizophrenia.

    International Registered Report Identifier (IRRID): RR2-10.1136/bmjopen-2020-046552

    JMIR Mhealth Uhealth 2025;13:e56185

    doi:10.2196/56185

    Keywords

    schizophrenia; psychosis; digital phenotyping; wrist wearables; mobile phone 


    Schizophrenia is a debilitating condition affecting at least 20 million individuals worldwide [1]. This lifelong condition [2] is frequently associated with impairments in personal, social, occupational, and other important areas of daily function [3]. Despite advancements in pharmacology and psychosocial interventions, recovery rates among patients have not improved [4], and the high rate of relapse places a heavy burden on patients and their caregivers [5].

    In present-day psychiatric care, the clinical status of patients is determined at periodic consultations via clinician interviews and self-reported questionnaires. The episodic nature of these visits and the removal of patients from their natural environment result in limited assessments that do not always reflect the state of patients in different environments and over varying periods of time [6]. Moreover, there is a risk of unreliable self-reporting in patients with schizophrenia, influenced by cognitive impairments, biases, and self-reporting confounders [7].

    Digital phenotyping may allow care teams to circumvent these limitations. Digital phenotyping is defined as the moment-by-moment quantification of the individual-level human phenotype in situ using data from smartphones and other personal digital devices [8,9]. Through personal digital devices, insightful and objective longitudinal data, such as on sleep and mobility, can be unobtrusively and continuously collected from patients [10].

    There has been growing interest in digital phenotyping in psychiatry [11,12]. People with schizophrenia have shown willingness to use smartphones to monitor their condition [13]. Furthermore, preliminary trials have shown that the long-term use of wearable technology is feasible and acceptable by people with schizophrenia and that paranoia toward devices was not a significant barrier [14-16].

    There is also some evidence that wearable and mobile devices can deliver clinically relevant information in real-world contexts. Movement data estimated from a 3-axis accelerometer were negatively correlated with negative symptoms assessed with the Positive and Negative Syndrome Scale (PANSS) in patients with schizophrenia [17]. The number of steps was negatively correlated with PANSS positive-factor, negative-factor, general subscales, and total score, in patients with schizophrenia in an inpatient setting [18]. Sleep duration estimated from mobile phone–based accelerometer data was moderately correlated with subject self-assessment of sleep duration [19]. Geolocation and motor activity inferred from gyroscopes and data from the ambient light sensor, accelerometer, and other sensors in smartphones were used as indices of negative symptoms [20,21]. Data from anonymized call and text message logs may be potential markers for social functioning, and data on screen touches, such as typing speed, may provide novel clinical data about a user’s cognitive functioning [9].

    Despite this evidence, past studies have noteworthy limitations, such as small sample sizes [17,22], nonclinical samples [23,24], being conducted in closed-ward settings [18], and having used made-for-research wearable devices [22], which would affect the generalizability of the research findings or scaling up to larger populations. To date, few studies have reported digital phenotyping measures collected with both low-cost, commercially available digital wearable devices and smartphones among patients with schizophrenia in a community setting.

    The aim of this study was to provide an initial examination of the associations between multimodal digital markers and clinical outcomes in patients with schizophrenia.


    Setting and Participants

    The Health Outcomes via Positive Engagement in Schizophrenia (HOPE-S) study is a prospective observational study conducted at the Institute of Mental Health, Singapore, with recently discharged individuals with schizophrenia spectrum disorders. Details of the study have been previously reported [25]. In brief, participants were followed up for up to 6 months.Clinical, cognitive, and functioning data were collected at intervals of every 6 weeks, and digital data from smartphones and wrist wearables were collected continuously over the follow-up period. The clinical, cognitive, and functioning data collected at baseline and the digital data collected in the first week after study enrollment were used for this study. The study aimed to examine the associations between clinical, cognitive, functioning, and digital data to develop prediction algorithms that might detect clinically significant outcomes. We recruited 100 English-speaking patients aged 21‐65 years between November 2019 and January 2023. One participant requested to be withdrawn from the study after completing the baseline assessment due to objections from family members. Of the remaining participants (n=99), 63 were diagnosed with schizophrenia, 12 with brief psychotic disorder, 11 with schizophreniform disorder, 5 with schizoaffective disorder, 5 with delusional disorder, and 3 with other specified psychotic disorder.

    Ethical Considerations

    Ethics approval was obtained from the National Healthcare Group’s Domain Specific Review Board, Singapore (2019/00720). Written informed consent was obtained from all participants after the study was fully explained. Data were anonymized and encrypted to protect patient privacy. Participants were reimbursed SGD 50 (US $37.19) on completion of the baseline visit for their time, inconvenience, and transportation costs.

    Clinical Symptom Severity Measures and Scales

    Assessment data from baseline were used for the present analyses. Symptom severity was assessed on the PANSS [26]. The PANSS is a 30-item clinical scale measuring positive symptoms, negative symptoms, and general psychopathology. PANSS factor scores, including positive (PANSS_Pos), negative (PANSS_Neg), cognitive/disorganization (PANSS_Cog/Dis), depression/anxiety (PANSS_Dep/Anx), and hostility (PANSS_Hos), were calculated in accordance with a previously validated factor structure [27].

    Negative symptoms were measured with the Brief Negative Symptom Scale (BNSS) [28], a well-established 13-item instrument designed to measure 5 negative symptom domains, including asociality, anhedonia, avolition, blunted affect, and alogia [29].

    Depressive symptoms were assessed on the Calgary Depression Scale for Schizophrenia (CDSS) [30], a 9-item scale widely used for assessing depression in schizophrenia that differentiates between depression symptoms and the negative and positive symptoms in schizophrenia.

    Neurocognition was assessed with the Brief Assessment of Cognition in Schizophrenia (BACS) [31], a cognitive battery that assesses attention, verbal and working memory, motor and processing speed, verbal fluency, and executive function. A composite z score was obtained and used for analysis [31]. The Social and Occupational Functioning Assessment Scale (SOFAS) [32] was used to evaluate the patients’ social and occupational functioning.

    Digital Data Collection and Preprocessing

    Digital data were collected from both wrist wearables and smartphones. Each participant was issued a Fitbit Charge 3 or 4 (Fitbit Inc) at the beginning of the study. Participants were required to wear the Fitbit every day during the study period, including when asleep. The Health Outcomes through Positive Engagement and Self-Empowerment (HOPES) app [33] was installed on a participant’s smartphone if the device used the Android operating system. In the circumstance that a participant’s phone was not compatible with the HOPES app, we provided a phone for use during the study period. In this case, study coordinators aided in the patient’s migration to the study phone, and the patient was instructed to use only the study phone as their primary mobile device during the study. Data collected from the wrist wearables and smartphones were automatically synchronized with the HOPES app and subsequently sent to the research premises for storage and further analysis. Data collected by the Fitbit included data from the heart rate monitor, pedometer (step counts), and sleep sensor, including sleep stages. Data collected from the smartphone included locational mobility features (through GPS), touchscreen tapping speed, and sociability indices from texts and calls (via both the mobile service provider and WhatsApp). Significant steps were taken to ensure data privacy, validate data quality, and engineer better features, which are covered in detail in a publication by Wang et al [33]. The subset of data analyzed in this paper is described in Multimedia Appendix 1. Table 1 summarizes the resulting digital measures used in our analyses.

    Table 1. Digital measures collected by Fitbit wearables and smartphones.
    Daily digital measuresData sourceDescription
    total_sleep_hrsWearableTotal hours of sleep, including naps but not including the “wake” stage (see below).
    sleep_hrs_wakeWearableTotal hours of sleep classified as the “wake” sleep stage by the Fitbit, estimated to be the time in bed not sleeping (ie, falling asleep, after waking up, and awaking during the night).
    sleep_hrs_lightWearableTotal hours of sleep classified as the “light” sleep stage by the Fitbit.
    sleep_hrs_deepWearableTotal hours of sleep classified as the “deep” sleep stage by the Fitbit.
    sleep_hrs_REMWearableTotal hours of sleep classified as the “REM” (rapid eye movement) sleep stage by the Fitbit.
    sleep efficiencyWearableSleep efficiency score from the Fitbit for the main sleep segment.
    HR_asleepWearableAverage of heart rate samples while asleep, as measured by the Fitbit heart and sleep monitors.
    total stepsWearableTotal steps taken, as measured by the Fitbit.
    dist_travelledSmartphoneTotal distance traveled, as computed by GPS processing routines.
    mode_intertap_distSmartphoneMode of a rolling 7-day distribution of intertap durations, excluding taps in games (a higher value indicates a slower tapping speed).
    total_msg_sentSmartphoneTotal messages (texts and images) sent via mobile service provider and WhatsApp.
    screen_timeSmartphoneTotal time of smartphone use as represented by the power state logs.

    For each daily digital measure, data collected during the 7 days following the baseline visit (not including the data on the day of the baseline visit) were used. The 1-week duration should account for any day-of-week effects in the digital measures. Digital markers were taken as the average of the observations for each digital measure over the week of interest, except for the mode_intertap_dist measure, for which the last valid (ie, nonmissing) value was taken since this is already a 7-day rolling window statistic. We required at least 4 valid observations in the week-long observation window to produce a nonnull marker.

    Successful data collection rates for the wearables as the fraction of hour-long windows with at least 1 successfully collected heart rate sample in the window of study (ie, the 7 days [168 hours] following the baseline visit) were calculated. A successful data collection rate for the smartphones was calculated similarly using the ambient light sensor. Note that these 2 data sources were continually sampled by each device regardless of user action, so long as the participant was compliant, that is, the wearable was correctly worn, and the phone was powered on with the app running in the background.

    Statistical Analysis

    Descriptive statistics on clinical measures, digital data measures, and successful data collection rates were analyzed. The successful data collection rates were compared between groups of participants who used study phones and their own phones with the Mann-Whitney U test. As an exploratory analysis to discover possible digital markers as factors associated with different clinical measures, a multiple linear regression was performed independently for each pair of digital markers (independent variable) and clinical measure (dependent variable). In each regression model, age was included as a covariate along with an interaction term with the digital measure, which has been reported as a significant factor in relapse behavior, psychosocial functioning, and psychotic symptoms [34,35]. Because a significant difference in digital data collection rates was noticed between participants using a study-provided phone and those using their own phone, we also included a variable indicating whether the participant was using a study-provided phone (along with its interaction with the digital measure) in all models regressing onto a digital measure from the smartphone. To easily compare the relative effect sizes between different measures, both the digital markers and clinical scales were standardized across the population; an intercept term was therefore not included in the models. Due to the particularly skewed distributions of the distance_travelled and total_msg_sent features, a log(1+x) transform was applied to these features before standardization. As this was an exploratory analysis attempting to uncover trends in the digital markers, multiple testing corrections were not used. A heat map was used to show effect sizes and visualize trends across digital markers and clinical scales.


    Sample Characteristics

    The demographic and clinical characteristics of the study sample are summarized in Table 2.

    Table 2. Demographic and clinical characteristics of the study sample (n=99).
    CharacteristicsValues
    Sex, n (%)
     Female57 (57.6)
     Male42 (42.4)
    Marital status, n (%)
     Single85 (85.9)
     Married7 (7.1)
     Separated or divorced7 (7.1)
    Ethnicity, n (%)
     Chinese73 (73.7)
     Malay17 (17.2)
     Indian6 (6.1)
     Other3 (3.0)
    Currently on antipsychotics, n (%)
     Yes97 (98.0)
     No2 (2.0)
    Age (years), mean (SD)29.88 (6.72)
    Years of education, mean (SD)14.07 (2.27)
    Duration of illness (years), mean (SD)2.61 (4.00)
    Antipsychotic dosea (mg), mean (SD)484.37 (375.55)
    PANSSbscore, mean (SD)
    Total51.11 (9.97)
     Positive8.49 (3.82)
     Negative13.20 (4.08)
     Cognitive/disorganization10.55 (2.50)
     Depression/anxiety8.85 (3.43)
     Hostility5.40 (1.89)
    BNSScscore, mean (SD)
    Total20.90 (9.30)
     Anhedonia5.51 (2.76)
     Asociality3.34 (1.55)
     Avolition3.30 (1.89)
     Blunted5.77 (3.89)
     Alogia2.82 (2.61)
    CDSSd total score, mean (SD)12.31 (3.55)
    SOFASe score, mean (SD)53.96 (11.85)
    BACSf score, mean (SD)−1.31 (1.36)

    aAntipsychotic dose was computed as total daily chlorpromazine equivalent (n=91).

    bPANSS: Positive and Negative Syndrome Scale.

    cBNSS: Brief Negative Symptom Scale.

    dCDSS: Calgary Depression Scale for Schizophrenia.

    eSOFAS: Social and Occupational Functioning Assessment Scale.

    fBACS: Brief Assessment of Cognition in Schizophrenia.

    Digital Data Collection Rate and Digital Data Profile

    In total, 41 and 58 participants were issued a Fitbit Charge 3 and Charge 4, respectively. Thirty-nine participants used a smartphone provided by the study at the baseline visit, while the rest used their own phone. The successful data collection rate in the week following the baseline visit was 91% (SD 22%) for the Fitbit and 82% (SD 32%) for the smartphone. Those using a study-provided phone had a significantly higher successful data completion rate than those using their own phone (mean 91% vs 77%; Mann-Whitney U test: P<.001). Sensor-specific successful data collection rates are also provided in Multimedia Appendix 2. The sample size of valid digital measures (n=99), along with the means and SDs, are presented in Table 3.

    Table 3. Summary of digital markers.
    Visit-level digital markersValues, mean (SD)
    Total_sleep_hrs (n=92)8.76 (2.13)
    Sleep hrs-wake (n=91)0.83 (0.30)
    Sleep hrs-light (n=91)5.32 (1.71)
    Sleep hrs-REM (n=91)1.68 (0.51)
    Sleep hrs-deep (n=91)1.23 (0.39)
    Sleep_efficiency (n=91)0.93 (0.03)
    Total_steps (n=96)9186.91 (5413.21)
    Distance_travelled (km) (n=87)27.64 (34.59)
    HR_asleep (bpm) (n=74)69.94 (10.06)
    Mode_intertap_dist (ms) (n=96)596.06 (389.08)
    Total_msg_sent (n=92)38.02 (65.16)
    Screen_time (hrs) (n=92)4.74 (3.64)

    As expected, several digital measures were correlated with one another (such as sleep stage measures). The correlation coefficients between digital measures are provided in Multimedia Appendix 3.

    Association Between Clinical Scales and Digital Markers

    The linear regression coefficients of the digital measures are visualized for the different clinical measures as a heat map (Figure 1) at the mean age level (29.88 years) and for participants using their own phone. The numbers in the cells are the value of the regression coefficient, which also determines the cell color and intensity.

    ‎
    Figure 1. Regression coefficients between clinical scales and digital markers. Notes: A log(1+x) transformation was applied to distance_travelled and total_msg_sent. All clinical scales and digital markers were standardized prior to the regression. The Social and Occupational Functioning Assessment Scale (SOFAS) score was inverted (denoted as inv_soafs_rating, calculated as 100 minus the SOFAS rating) in the analyses to maintain a consistent direction among the scales and the digital markers, which aids visualization and interpretation. Analogously, the Brief Assessment of Cognition in Schizophrenia composite score was inverted (denoted as neg_bacs_comp) by taking the negative of its value. Ordinary least squares: *POLS<.05, **POLS<.01.

    Starting with the symptoms and their associated digital biomarkers, first, we found that negative symptoms as assessed by both the BNSS and PANSS-negative were associated with the most digital measures (8 of the 12 being studied). Experiential deficits, that is, anhedonia and avolition, tended to be associated with less distance traveled and more hours spent awake in bed; fewer messages sent, a digital marker of sociability, tended to be associated with asociality. Expressive deficits, that is, affect blunting and alogia, tended to be negatively associated with steps taken. Alogia was additionally associated with sleep features such as total sleep hours and hours in light and rapid eye movement (REM) sleep. Second, PANSS-positive and SOFAS scores were associated with at least 3 different digital measures. Third, cognition and cognitive/disorganization symptoms were associated with 1 or 2 digital measures.

    Among the digital measures, heart rate measured when asleep was associated with the most clinical measures. Specifically, a higher heart rate was associated with greater severity scores on PANSS-total and its factors (ie, positive, negative, and cognitive/disorganization) and BNSS avolition and asociality, as well as poorer functioning and cognitive performance. Time awake in bed and distance traveled were associated with mostly similar clinical measures, that is, negative symptoms, in particular anhedonia, avolition, and functioning. The only digital measure not significantly associated with a clinical measure was screen time.


    Principal Findings

    In this study, we examined the associations between digital markers (collected via commercially available wrist-wearable devices and smartphones) and a comprehensive range of clinical assessments in a sample of 99 recently discharged patients with schizophrenia spectrum disorders. Notable patterns of associations for digital measures were observed with psychiatric symptoms, specifically negative symptoms, functioning, and cognitive performance.

    Longer total sleep time was positively associated with negative symptom domains; sleep efficiency was negatively associated with BNSS avolition and the PANSS cognitive/disorganization score. These results are in line with previous studies that reported associations between negative symptom severity and sleep disturbances (such as hypersomnia and sleep satisfaction) collected through self-reporting [36,37] and as measured by actigraphy [22,38]. One study also found no correlation between sleep markers collected via actigraphy and negative symptoms assessed via the PANSS-negative symptom subscale [39]. In addition, whether and how each sleep stage is associated with symptoms in schizophrenia remains inconclusive. A previous study reported that REM sleep duration was correlated with the Brief Psychiatric Rating Scale (BPRS) total score but not with the constituent negative symptom score [40]; in contrast, this study found no association between REM sleep duration and PANSS-total score but found a positive association with alogia. Use of nonspecific negative symptom assessment scales such as PANSS and BPRS, variations in sleep instruments, and other clinical factors, such as duration of illness and medication status [41], likely contributed to the mixed results in the literature.

    Several types of passive digital phenotyping measures are associated with negative symptoms, such as phone use data (call/text logs, screen time), GPS data (distance traveled), sleep data (total sleep duration), and physical activity (step count). These findings support the use of passively collected digital phenotyping data as objective measures of negative symptoms and functional outcomes. These measures also provide longitudinal data that are usually not available directly to clinicians. Our study found a negative association between physical activity (as measured by step count) and negative symptom severity, particularly BNSS avolition, blunted affect, and alogia. This is in line with previous research that showed associations between decreased activity or sedentary behavior and negative symptoms in schizophrenia [17,18,22,42-44].

    Distance traveled, messages sent and touchscreen tapping speed yielded the strongest signals associated with negative symptoms, specifically experiential deficits such as avolition, anhedonia and asociality. These results are consistent with previous studies that reported greater GPS mobility was associated with lower negative symptom severity and better community functioning [20,45]. Social behaviors (such as duration of outgoing calls and number of text messages) have been found to be predictive of psychotic symptoms and relapse in schizophrenia [46,47]. Typing-related data have been reported to be associated with mood symptoms and cognition in people without psychosis [48-50], but no such association was found in our study. The current study appears to be the first to report significant associations between negative symptoms, specifically asociality, and a messaging log feature (ie, the number of messages sent out) as well as touchscreen tapping speed.

    We found that increased heart rate while asleep was associated with increases in positive psychopathology and worse cognitive performance. The relationship between increased heart rate and a greater severity of positive symptoms has often been documented [51-53], but only one previous study reported this relationship for heart rate during sleep, supporting our findings [54]. These associations likely indicate altered autonomic functioning, as indexed by increased heart rate, in response to stressors such as manifestations of psychotic symptoms [51,55]. Similarly, altered autonomic functioning may affect cognition. While no studies have reported the association of increased heart rate during sleep with worsening cognition, reduced heart rate variability has been linked to worse cognitive scores in schizophrenia [56].

    The successful data collection rates in this study were high—about 91% for the Fitbit devices and 82% for the smartphones, which is comparable to the compliance rates in previous studies [15,19]. The good data collection rates support the potential of wrist wearables and smartphones as longitudinal passive monitoring devices in people with schizophrenia. It should be noted that the enrolled sample was relatively young, with a mean age of 30 years, and this group of people may be more accepting of digital devices. The digital data collection rate from the smartphones was impacted by both the compliance of participants and technical factors, such as proper device functioning and syncing with the mobile app. Some smartphone models automatically suppress all app activity during the night to save battery power, preventing complete data collection from the phone. In addition, some participants were provided with a phone for the study because their personal phone was incompatible with the HOPES app, and it was often the case that the participant’s phone was too old. The above 2 points could explain why the data collection rate was higher with study-provided phones, as these (newer) phones could have exhibited better functioning.

    Although there is a considerable amount of literature reporting associations between digital markers and clinical symptoms in populations of outpatients with schizophrenia, the majority collected digital data via only smartphones or actigraphy devices. Despite the expanding interest in using commercially available wearables in health care [57], studies using these devices in the medical field are still few. This study contributes such an investigation with a large clinical sample. Notably, the data reported in this study were not affected by Singapore’s COVID-19 movement control measures: none of the 1-week digital data collection periods under study overlapped with the movement control period.

    Limitations

    There are some limitations to this study to be considered. The participants were relatively young (mean age: 30 years) and mildly ill (PANSS-total score: 51) [58] with a short duration of illness (2.6 years), which limits the generalizability of the results. Associations in older, chronic, and more severely ill patients might differ and should be further investigated. Due to the nature of this cross-sectional study, no causal relationships could be inferred. In addition, all participants included in this study were taking psychotropic medications with varying types and dosages, which may have had differential effects on cardiac function [59] and sleep [60], warranting further examination. Participants received financial incentives at timed intervals in the study, which might have positively affected adherence and data acquisition. Similar studies, however, noted that the majority of participants expressed a desire to continue using the wrist wearable without incentive after the study [19]. The longer-term use of wearable devices by individuals with schizophrenia should be further studied. Finally, because the regression analyses were exploratory in nature, and we aimed to detect all potential indicators of symptoms and functioning that could be useful in future studies, corrections for multiple testing were not applied before reporting significant associations.

    Despite the limitations of the study, the patterns highlighted in the findings may guide future directions for research. The reported associations suggest that unobtrusively and passively collected digital phenotyping markers hold promise for continuous monitoring of clinical symptoms in people with schizophrenia between clinical visits. Digital phenotyping–enhanced care can improve the identification of patients in need of attention and may allow health care providers to delivery timely interventions [61].

    Conclusion

    Our study found notable patterns of associations between a range of digital biomarkers and clinically important outcome measures in a community sample of individuals with schizophrenia spectrum disorders. These results support the use of commercial wrist wearables and smartphones in continuous but passive monitoring of the health status of individuals with schizophrenia. Future studies should explore longer-term digital phenotyping observations in naturalistic settings and their utility in predicting clinical events.

    Acknowledgments

    The authors would like to express their sincere gratitude to the participants for their valuable contributions and to the clinicians who helped in study recruitment and assessments. We would like to thank Dr Gurpreet Rekhi, who trained and supervised the research psychologists in conducting the clinical assessments. The study was supported by the Ministry of Health Office for Healthcare Transformation and the Ministry of Health National Medical Research Council Centre Grant (NMRC/CG/M002/2017_IMH). JL is supported by the Ministry of Health National Medical Research Council Clinician Scientist Award (NMRC/CSAINV17nov005). The funders played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript.

    Data Availability

    The datasets generated and/or analyzed during the current study are not publicly available, because the participants in this study did not give written consent for their data to be shared publicly.

    Authors' Contributions

    JL, RJTM, ZY, and CH designed the study. RJTM and JL obtained the funding. ZY, NAAR, and TB managed the study. ZY, AS, NAAR, SZ, YFQ, SB, KWL, CT, SV, and JL supported recruitment and data collection. CH, ZY, and XW analyzed the data. CH, ZY, JL, and RJTM interpreted the results. ZY and CH drafted the manuscript. All authors revised and approved the final version of the manuscript.

    Conflicts of Interest

    None declared.

    Multimedia Appendix 1

    Data collection and feature processing in the Health Outcomes through Positive Engagement and Self-Empowerment (HOPES) platform.

    DOCX File, 19 KB
    Multimedia Appendix 2

    Sensor-specific successful data collection rates.

    DOCX File, 16 KB
    Multimedia Appendix 3

    Correlations between digital measures.

    DOCX File, 176 KB
    1. GBD 2017 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet. Nov 2018;392(10159):1789-1858. [CrossRef] [Medline]
    2. Lee J, Remington G. Schizophrenia as a lifelong illness: implications for care. Ann Acad Med Singap. Jan 2015;44(1):26-28. [Medline]
    3. Bechi M, Bosia M, Spangaro M, et al. Exploring functioning in schizophrenia: predictors of functional capacity and real-world behaviour. Psychiatry Res. May 2017;251:118-124. [CrossRef] [Medline]
    4. Jääskeläinen E, Juola P, Hirvonen N, et al. A systematic review and meta-analysis of recovery in schizophrenia. Schizophr Bull. Nov 2013;39(6):1296-1306. [CrossRef] [Medline]
    5. Pennington M, McCrone P. The cost of relapse in schizophrenia. Pharmacoeconomics. Sep 2017;35(9):921-936. [CrossRef] [Medline]
    6. Cohen AS, Schwartz E, Le T, et al. Validating digital phenotyping technologies for clinical use: the critical importance of “resolution”. World Psychiatry. Feb 2020;19(1):114-115. [CrossRef] [Medline]
    7. Durand D, Strassnig M, Sabbag S, et al. Factors influencing self-assessment of cognition and functioning in schizophrenia: implications for treatment studies. Eur Neuropsychopharmacol. Feb 2015;25(2):185-191. [CrossRef] [Medline]
    8. Insel TR. Digital phenotyping: technology for a new science of behavior. JAMA. Oct 3, 2017;318(13):1215-1216. [CrossRef] [Medline]
    9. Torous J, Kiang MV, Lorme J, Onnela JP. New tools for new research in psychiatry: a scalable and customizable platform to empower data driven smartphone research. JMIR Ment Health. May 5, 2016;3(2):e16. [CrossRef] [Medline]
    10. Torous J, Keshavan M. A new window into psychosis: The rise digital phenotyping, smartphone assessment, and mobile monitoring. Schizophr Res. Jul 2018;197:67-68. [CrossRef] [Medline]
    11. Griffin B, Saunders KEA. Smartphones and wearables as a method for understanding symptom mechanisms. Front Psychiatry. 2019;10:949. [CrossRef] [Medline]
    12. Seppälä J, De Vita I, Jämsä T, et al. Mobile phone and wearable sensor-based mHealth approaches for psychiatric disorders and symptoms: systematic review. JMIR Ment Health. Feb 20, 2019;6(2):e9819. [CrossRef] [Medline]
    13. Beard C, Silverman AL, Forgeard M, Wilmer MT, Torous J, Björgvinsson T. Smartphone, social media, and mental health app use in an acute transdiagnostic psychiatric sample. JMIR Mhealth Uhealth. Jun 7, 2019;7(6):e13364. [CrossRef] [Medline]
    14. Cella M, He Z, Killikelly C, Okruszek Ł, Lewis S, Wykes T. Blending active and passive digital technology methods to improve symptom monitoring in early psychosis. Early Interv Psychiatry. Oct 2019;13(5):1271-1275. [CrossRef] [Medline]
    15. Meyer N, Kerz M, Folarin A, et al. Capturing rest-activity profiles in schizophrenia using wearable and mobile technologies: development, implementation, feasibility, and acceptability of a remote monitoring platform. JMIR Mhealth Uhealth. Oct 30, 2018;6(10):e188. [CrossRef] [Medline]
    16. Kerz M, Folarin A, Meyer N, Begale M, MacCabe J, Dobson RJ. SleepSight: a wearables-based relapse prevention system for schizophrenia. In: Proceedings of the 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing: Adjunct. Association for Computing Machinery; 2016:113-116. [CrossRef]
    17. Cella M, Okruszek Ł, Lawrence M, Zarlenga V, He Z, Wykes T. Using wearable technology to detect the autonomic signature of illness severity in schizophrenia. Schizophr Res. May 2018;195:537-542. [CrossRef] [Medline]
    18. Shin S, Yeom CW, Shin C, et al. Activity monitoring using a mHealth device and correlations with psychopathology in patients with chronic schizophrenia. Psychiatry Res. Dec 30, 2016;246:712-718. [CrossRef] [Medline]
    19. Staples P, Torous J, Barnett I, et al. A comparison of passive and active estimates of sleep in a cohort with schizophrenia. NPJ Schizophr. Oct 16, 2017;3(1):37. [CrossRef] [Medline]
    20. Raugh IM, James SH, Gonzalez CM, et al. Geolocation as a digital phenotyping measure of negative symptoms and functional outcome. Schizophr Bull. Dec 1, 2020;46(6):1596-1607. [CrossRef] [Medline]
    21. Strauss GP, Raugh IM, Zhang L, et al. Validation of accelerometry as a digital phenotyping measure of negative symptoms in schizophrenia. Schizophrenia (Heidelb). Apr 15, 2022;8(1):37. [CrossRef] [Medline]
    22. Kluge A, Kirschner M, Hager OM, et al. Combining actigraphy, ecological momentary assessment and neuroimaging to study apathy in patients with schizophrenia. Schizophr Res. May 2018;195:176-182. [CrossRef] [Medline]
    23. Ben-Zeev D, Scherer EA, Wang R, Xie H, Campbell AT. Next-generation psychiatric assessment: using smartphone sensors to monitor behavior and mental health. Psychiatr Rehabil J. Sep 2015;38(3):218-226. [CrossRef] [Medline]
    24. Wang R, Chen F, Chen Z, Li T, Harari G, Tignor S, et al. StudentLife: assessing mental health, academic performance and behavioral trends of college students using smartphones. In: UbiComp ’14: Proceedings of the 2014 ACM International Joint Conference on Pervasive and Ubiquitous Computing. Association for Computing Machinery; 2014:3-14. [CrossRef]
    25. Abdul Rashid NA, Martanto W, Yang Z, et al. Evaluating the utility of digital phenotyping to predict health outcomes in schizophrenia: protocol for the HOPE-S observational study. BMJ Open. Oct 20, 2021;11(10):e046552. [CrossRef] [Medline]
    26. Kay SR, Fiszbein A, Opler LA. The Positive and Negative Syndrome Scale (PANSS) for schizophrenia. Schizophr Bull. Jan 1, 1987;13(2):261-276. [CrossRef]
    27. Lim K, Peh OH, Yang Z, et al. Large-scale evaluation of the Positive and Negative Syndrome Scale (PANSS) symptom architecture in schizophrenia. Asian J Psychiatr. Aug 2021;62:102732. [CrossRef] [Medline]
    28. Kirkpatrick B, Strauss GP, Nguyen L, et al. The Brief Negative Symptom Scale: psychometric properties. Schizophr Bull. Mar 2011;37(2):300-305. [CrossRef] [Medline]
    29. Ang MS, Rekhi G, Lee J. Validation of the Brief Negative Symptom Scale and its association with functioning. Schizophr Res. Jun 2019;208:97-104. [CrossRef] [Medline]
    30. Addington D, Addington J, Maticka-Tyndale E, Joyce J. Reliability and validity of a depression rating scale for schizophrenics. Schizophr Res. Mar 1992;6(3):201-208. [CrossRef] [Medline]
    31. Keefe RSE, Goldberg TE, Harvey PD, Gold JM, Poe MP, Coughenour L. The Brief Assessment of Cognition in Schizophrenia: reliability, sensitivity, and comparison with a standard neurocognitive battery. Schizophr Res. Jun 1, 2004;68(2-3):283-297. [CrossRef] [Medline]
    32. Rybarczyk B. Social and Occupational Functioning Assessment Scale (SOFAS). In: Kreutzer JS, DeLuca J, Caplan B, editors. Encyclopedia of Clinical Neuropsychology. Springer; 2011:2313. [CrossRef]
    33. Wang X, Vouk N, Heaukulani C, et al. HOPES: an integrative digital phenotyping platform for data collection, monitoring, and machine learning. J Med Internet Res. Mar 15, 2021;23(3):e23984. [CrossRef] [Medline]
    34. Alphs L, Nasrallah HA, Bossie CA, et al. Factors associated with relapse in schizophrenia despite adherence to long-acting injectable antipsychotic therapy. Int Clin Psychopharmacol. Jul 2016;31(4):202-209. [CrossRef] [Medline]
    35. Jeste DV, Maglione JE. Treating older adults with schizophrenia: challenges and opportunities. Schizophr Bull. Sep 2013;39(5):966-968. [CrossRef] [Medline]
    36. de Crom SAM, Haan LD, Schirmbeck F, GROUP investigators. The association between sleep disturbances and negative symptom severity in patients with non-affective psychotic disorders, unaffected siblings and healthy controls. Psychiatry Res. Mar 2021;297:113728. [CrossRef] [Medline]
    37. Blanchard JJ, Andrea A, Orth RD, Savage C, Bennett ME. Sleep disturbance and sleep-related impairment in psychotic disorders are related to both positive and negative symptoms. Psychiatry Res. Feb 7, 2020;286:112857. [CrossRef] [Medline]
    38. Docx L, Sabbe B, Provinciael P, Merckx N, Morrens M. Quantitative psychomotor dysfunction in schizophrenia: a loss of drive, impaired movement execution or both? Neuropsychobiology. 2013;68(4):221-227. [CrossRef] [Medline]
    39. Chung KF, Poon Y, Ng TK, Kan CK. Correlates of sleep irregularity in schizophrenia. Psychiatry Res. Dec 2018;270:705-714. [CrossRef] [Medline]
    40. Poulin J, Daoust AM, Forest G, Stip E, Godbout R. Sleep architecture and its clinical correlates in first episode and neuroleptic-naive patients with schizophrenia. Schizophr Res. Jul 1, 2003;62(1-2):147-153. [CrossRef] [Medline]
    41. Chan MS, Chung KF, Yung KP, Yeung WF. Sleep in schizophrenia: a systematic review and meta-analysis of polysomnographic findings in case-control studies. Sleep Med Rev. Apr 2017;32:69-84. [CrossRef] [Medline]
    42. Abdul Rashid NA, Nurjono M, Lee J. Clinical determinants of physical activity and sedentary behaviour in individuals with schizophrenia. Asian J Psychiatr. Dec 2019;46:62-67. [CrossRef] [Medline]
    43. Wichniak A, Skowerska A, Chojnacka-Wójtowicz J, et al. Actigraphic monitoring of activity and rest in schizophrenic patients treated with olanzapine or risperidone. J Psychiatr Res. Oct 2011;45(10):1381-1386. [CrossRef] [Medline]
    44. Walther S, Ramseyer F, Horn H, Strik W, Tschacher W. Less structured movement patterns predict severity of positive syndrome, excitement, and disorganization. Schizophr Bull. May 2014;40(3):585-591. [CrossRef] [Medline]
    45. Depp CA, Bashem J, Moore RC, et al. GPS mobility as a digital biomarker of negative symptoms in schizophrenia: a case control study. NPJ Digit Med. 2019;2:108. [CrossRef] [Medline]
    46. He-Yueya J, Buck B, Campbell A, et al. Assessing the relationship between routine and schizophrenia symptoms with passively sensed measures of behavioral stability. NPJ Schizophr. Nov 23, 2020;6(1):35. [CrossRef] [Medline]
    47. Buck B, Scherer E, Brian R, et al. Relationships between smartphone social behavior and relapse in schizophrenia: a preliminary report. Schizophr Res. Jun 2019;208:167-172. [CrossRef] [Medline]
    48. Mastoras RE, Iakovakis D, Hadjidimitriou S, et al. Touchscreen typing pattern analysis for remote detection of the depressive tendency. Sci Rep. Sep 16, 2019;9(1):13414. [CrossRef] [Medline]
    49. Ross MK, Demos AP, Zulueta J, et al. Naturalistic smartphone keyboard typing reflects processing speed and executive function. Brain Behav. Nov 2021;11(11):e2363. [CrossRef] [Medline]
    50. Kapsecker M, Osterlehner S, Jonas SM. Analysis of mobile typing characteristics in the light of cognition. In: 2022 IEEE International Conference on Digital Health (ICDH). IEEE; 2022. [CrossRef]
    51. Bär KJ, Wernich K, Boettger S, et al. Relationship between cardiovagal modulation and psychotic state in patients with paranoid schizophrenia. Psychiatry Res. Jan 15, 2008;157(1-3):255-257. [CrossRef] [Medline]
    52. Chang HA, Chang CC, Tzeng NS, Kuo TBJ, Lu RB, Huang SY. Cardiac autonomic dysregulation in acute schizophrenia. Acta Neuropsychiatr. Jun 2013;25(3):155-164. [CrossRef] [Medline]
    53. Chung MS, Yang AC, Lin YC, et al. Association of altered cardiac autonomic function with psychopathology and metabolic profiles in schizophrenia. Psychiatry Res. Dec 30, 2013;210(3):710-715. [CrossRef] [Medline]
    54. Kalisperakis E, Karantinos T, Lazaridi M, et al. Smartwatch digital phenotypes predict positive and negative symptom variation in a longitudinal monitoring study of patients with psychotic disorders. Front Psychiatry. 2023;14:1024965. [CrossRef] [Medline]
    55. Stogios N, Gdanski A, Gerretsen P, et al. Autonomic nervous system dysfunction in schizophrenia: impact on cognitive and metabolic health. NPJ Schizophr. Apr 26, 2021;7(1):22. [CrossRef] [Medline]
    56. Haigh SM, Walford TP, Brosseau P. Heart rate variability in schizophrenia and autism. Front Psychiatry. 2021;12:760396. [CrossRef] [Medline]
    57. Smuck M, Odonkor CA, Wilt JK, Schmidt N, Swiernik MA. The emerging clinical role of wearables: factors for successful implementation in healthcare. NPJ Digit Med. Mar 10, 2021;4(1):45. [CrossRef] [Medline]
    58. Leucht S, Kane JM, Kissling W, Hamann J, Etschel E, Engel RR. What does the PANSS mean? Schizophr Res. Nov 15, 2005;79(2-3):231-238. [CrossRef] [Medline]
    59. Huhn M, Arndt T, Schneider-Thoma J, Leucht S. Effects of antipsychotics on heart rate in treatment of schizophrenia: a systematic review and meta-analysis. Ther Adv Psychopharmacol. 2022;12:20451253221097261. [CrossRef] [Medline]
    60. Cohrs S. Sleep disturbances in patients with schizophrenia: impact and effect of antipsychotics. CNS Drugs. 2008;22(11):939-962. [CrossRef] [Medline]
    61. Stafford MR, Jackson H, Mayo-Wilson E, Morrison AP, Kendall T. Early interventions to prevent psychosis: systematic review and meta-analysis. BMJ. Jan 18, 2013;346:f185. [CrossRef] [Medline]

    ‎
    BACS: Brief Assessment of Cognition in Schizophrenia
    BNSS: Brief Negative Symptom Scale
    BPRS: Brief Psychiatric Rating Scale
    CDSS: Calgary Depression Scale for Schizophrenia
    HOPE-S: Health Outcomes via Positive Engagement in Schizophrenia
    HOPES: Health Outcomes through Positive Engagement and Self-Empowerment
    PANSS: Positive and Negative Syndrome Scale
    REM: rapid eye movement
    SOFAS: Social and Occupational Functioning Assessment Scale

    Edited by Lorraine Buis; submitted 09.01.24; peer-reviewed by Lulu Yang, Marcin Straczkiewicz, Philip Harvey; final revised version received 25.09.24; accepted 02.12.24; published 05.02.25.

    Copyright

    © Zixu Yang, Creighton Heaukulani, Amelia Sim, Thisum Buddhika, Nur Amirah Abdul Rashid, Xuancong Wang, Shushan Zheng, Yue Feng Quek, Sutapa Basu, Kok Wei Lee, Charmaine Tang, Swapna Verma, Robert J T Morris, Jimmy Lee. Originally published in JMIR mHealth and uHealth (https://mhealth.jmir.org), 5.2.2025.

    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.