Skip to main content
. 2023 Dec 13;25:e46778. doi: 10.2196/46778

Table 3.

Schizophrenia: retrieved studies and their main outcomes.

Study Country (region) Data Psychiatric disorder Symptoms investigated Assessment technology Sample, n Data collection time Statistics Synthesis of main results
Abbas et al [10], 2021 New York, United States 2021 Schizophrenia Euclidean distance of head movements frame by frame Passive (smartphone): front-facing camera of the smartphone

  • Schizophrenia: 18

  • Controls: 9

  • Period of collection: 2 weeks

  • Frequency of collection: once a day

  • Rate of head movement with and without schizophrenia (1.48 mm/frame vs 2.50 mm/frame; P=.01)

  • Head movement predictor of schizophrenia diagnosis (P=.02)

  • Head movement and negative symptoms (t=−2.245; P=.04)

 Smartphone-based remote assessments were able to capture meaningful visual behaviors for objective measurement of head movements of patients with schizophrenia, demonstrating that it is possible to quantify the severity of their symptoms, particularly negative symptoms, by head movements.
Abbas et al [11], 2022 New York, United States 2019 Schizophrenia Facial and vocal characteristics, including facial expressivity, vocal acoustics, and speech prevalence

  • Passive (smartphone): front-facing smartphone camera and microphone

  • Active: PANSSa

 20

  • Period of collection: 14 days

  • Frequency of collection: passive (continuously) and active (twice)

  • Significant correlations between vocal markers and N total were fundamental frequency mean (r=−0.64; adjusted P=.02), vocal jitter (r=0.56; adjusted P=.02), and harmonics to noise ratio (r=−0.61; adjusted P=.02).

  • Significant correlations between free behavior in response to images and N total: fundamental frequency mean (r=−0.61; adjusted P=.04), harmonics to noise ratio (r=−0.58; adjusted P=.03), and speech prevalence (r=−0.57; adjusted P=.03).

 Facial and vocal measurements collected remotely in patients with schizophrenia via smartphones in response to automated task prompts demonstrated accuracy as markers of schizophrenia symptom severity.
Adler et al [12], 2020 United States 2020 Schizophrenia Environmental data, sleep, SMS text messages, location, calls, app use, screen interaction

  • Passive (smartphone): app CrossCheck

  • Active: EMAb

 60

  • Period of collection: 12 months

  • Frequency of collection: passive (continuously) and active (every 2-3 days)

  • Model sensitivity: 0.25 (IQR 0.15-1.00)

  • Model specificity: 0.88 (IQR 0.14-0.96)

  • A median 108% increase in behavioral anomalies near relapse


 Using digital tools in mental health, it is possible to predict relapse in patients with schizophrenia.
Barnett et al [13], 2018 United States 2018 Schizophrenia Mobility, sociality, mood

  • Passive (smartphone): GPS, accelerometer, anonymous call and SMS text message logs, screen on/off time, phone charge status

  • Active: biweekly surveys

 17

  • Period of collection: 3 months

  • Frequency of collection: passive (continuously) and active (every 2 weeks)

  • The rate of behavioral abnormalities detected in the 2 weeks before relapse was 71% higher than during other periods.

 Results showed that, with proper supporting instrumentation, smartphones can be used as research tools within mental health.
Henson et al [14], 2020 Boston, MA, United States 2020 Schizophrenia Sociality, mobility, sleep, anxiety and depressive symptoms, psychotic symptoms

  • Passive (smartphone): GPS, accelerometer, screen on/off and call/SMS logs

  • Active: PHQ-9c, GAD-7d

 92

  • Period of collection: 3 months

  • Frequency of collection: passive (continuously) and active (10 times in 3 months)

  • Clinical population: passive data features and symptoms—Spearman ρ ranged from −0.23 to −0.30, P<.001.

  • Healthy controls: passive data features and symptoms—Spearman ρ ranged from 0.20 to 0.44, P<.05

 The results suggest that digital phenotyping in schizophrenia may offer clinically relevant information for understanding how daily routines influence symptoms.
Henson et al [15], 2021 Boston, MA, United States 2021 Schizophrenia Mobility, sociality, sleep, cognition

  • Passive (smartphone): GPS, calls, messages, accelerometer, screen use

  • Active: PHQ-9, GAD-7, PANSS, CGI (clinical global impression)

  • Schizophrenia=83

  • Controls=43

  • Period of collection: 2 years

  • Frequency of collection: passive (twice each day or 5 times each week) and active (every 1-3 months)

 Sensitivity=89% and specificity=75% in predicting symptom relapse The model used confirmed the potential and clinical utility of longitudinal collection of symptomatologic and behavioral data.
Ranjan et al [16], 2022 Boston, United States 2019-2021 Schizophrenia Mood, sleep, and psychosis symptoms Passive (smartphone): sensors (GPS, accelerometer, screen time, call and text logs)
Active: PANSS and EMA		 86

  • Period of collection: 21 months

  • Frequency of collection: passive (continuously) and active (5 clinical visits and undefined amount of survey completion)

 Correlation between gold standard and app-based self-report symptom (r=0.8, P=10−11 for mood and r=0.78, P=10−12 for anxiety) The intraindividual symptom correlations and the stratification of symptoms by home time highlight the utility of digital phenotyping methods as a diagnostic tool, as well as the potential for personalized psychiatric treatment based on these data.
Strauss et al [17], 2022 United States 2022 Schizophrenia Negative psychosis symptoms Passive (wearables through smartphone app): accelerometry
Passive (smartphone): accelerometry
Active: SCID-5e, BNSSf, PANSS, and LOFg

  • Schizophrenia=50

  • Control=70

  • Period of collection: 6 days

  • Frequency of collection: passive (continuously) and active (8 daily surveys)

 Group differences in accelerometry:
Phone ACLh:

  • Mean: F1, 118=17.75, P<.001

  • SD: F1, 118=23.64, P<.001

Band ACL:

  • Mean: F1, 64=.06, P=.8

  • SD: F1, 64=.60, P=.44

  • Activity index: F1, 64=0.47, P=.5

Convergent validity:
ACL band mean and

  • BNSS anhedonia=−0.45, P<.05

  • BNSS avolition=−0.41, P<.05

  • BNSS blunted affect=−0.4, P<.05

  • BNSS alogia=−0.4, P<.05

 Accelerometry is a valid objective measure of negative symptoms that may complement traditional approaches to assess the construct using clinical rating scales.
Wang et al [18], 2016 Heidelberg, Germany 2016 Schizophrenia Location, sleep, sociality, smartphone use, and environmental conditions Passive (smartphone): CrossCheck App
Active: EMA		 21

  • Period of collection: 12 months

  • Frequency of collection: passive (continuously) and active (3 times a week)

 Mean prediction model error of 7.6% of the score range We show that by leveraging knowledge from a population with schizophrenia, it is possible to train accurate personalized models that require less individual-specific data to rapidly adapt to new users.

aPANSS: Positive and Negative Syndrome Scale.

bEMA: Ecological Momentary Assessment

cPHQ-9: Patient Health Questionnaire-9.

dGAD-7: Generalized Anxiety Disorder-7.

eSCID-5: Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition.

fBNSS: Brief Negative Symptom Scale.

gLOF: Level of Functioning Scale.

hACL: Accelerometry.