Table 1.
Summary of original papers.
| Source | Sample description | Collected data | Related clinical measures | Results |
| Ben-Zeev et al [49] | 47 healthy subjects | GPS, accelerometer, gyroscope, microphone, and light sensor | PHQ-9a, PSSb, and revised UCLAc loneliness scale | Speech duration, sleep duration, and geospatial activity relate to PHQ-9; kinesthetic activity relates to UCLA loneliness scale. |
| Osmani V et al [50] | 9 subjects with bipolar disorders | Accelerometer and gyroscope | HAMDd and YMRSe | Psychiatric assessment scores relate to physical activity level at specific time intervals of the day. |
| Chow P et al [51] | 72 healthy subjects | GPS | SIASf and DASS-21g | Social anxiety and depression relate to time spent at home in specific time intervals of the day. |
| Boukhechba et al [52] | 54 healthy subjects | GPS, phone calls, and messages | SIAS | Social anxiety relates to limited social life and reduced mobility. |
| Staples et al [53] | 17 subjects with schizophrenia | Accelerometer and gyroscope | PSQIh | Moderate correlation between sleep estimate and PSQI. |
| Sano et al [54] | 66 healthy subjects | Accelerometer, gyroscope, skin temperature, skin conductance, phone calls, messages, and screen on/off | PSQI, Big Five Inventory Personality Test, MEQi, PSS, and MCS for mental healthj | PSQI and stress relate to phone usage. |
| Sano et al [55] | 18 healthy subjects | GPS, accelerometer, gyroscope, skin conductance, phone calls, messages, and screen on/off | PSS, PSQI, and Big Five Inventory Personality Test | Stress relates to phone usage and physical activities at specific time intervals of the day. |
| Stutz et al [56] | 15 healthy subjects | Accelerometer, gyroscope, light, app usage, and screen on/off | PSS | PSS relates mainly to phone usage. |
| Difrancesco et al [57] | 7 subjects with schizophrenia | GPS | Birchwood’s Social Functioning Scale | Locations detected through GPS relate well to the activities identified in the social functioning scale. |
| Osmani V [58] | 12 subjects with bipolar disorders | GPS, accelerometer, gyroscope, and microphone | Mental scale (not specifically defined) | Physical activity and voice features relate to the patient’s state. |
| Renn B et al [59] | 600 subjects with depression | GPS | PHQ-2k | Limited association between mobility and depressive symptoms rating. |
| Mehrotra et all [60] | 25 healthy subjects | Phone notification management (eg, clicks, decision, and response time), phone calls, and app usage | PHQ-8l | Moderate correlation between depression state and notification management as well as phone and app usage in a 14-day period; limited correlation on shorter periods of time. |
| Grunerbl et al [61] | 10 subjects with bipolar disorders | GPS, accelerometer, gyroscope, microphone, and phone calls | HAMD and YMRS | Good relationship between sensor data and the patient’s state. |
| Saeb et al [62] | 28 healthy subjects | GPS and phone usage | PHQ-9 | Good relationship between phone usage (ie, calls and duration) and depression symptoms as well as GPS processed data and depression symptoms. |
| Guidi et al [63] | 1 patient with bipolar disorder | Microphone | QIDm and YMRS | No clear relationship between voice features and clinical assessment. |
| Beiwinkel et al [64] | 13 subjects with bipolar disorders | GPS, phone calls, and messages | HAMD and YMRS | Phone usage relates positively to depression state while activity relates negatively to manic symptoms. |
| Wahle et al [65] | 126 healthy subjects | GPS, accelerometer, and phone usage | PHQ-9 | Depression symptoms relate to mobile phone extracted features. |
| Shin et al [66] | 61 patients with schizophrenia, DSM-IVn | Fitbit (ie, activity tracker) | PANSSo | Psychiatric symptoms relate to lower activity level. |
| Palmius et al [67] | 29 subjects with bipolar disorders and 20 controls | GPS | QID | Location recordings relate to depressive episodes. |
| Abrantes et al [68] | 20 subjects with alcohol use disorders | Fitbit (ie, activity tracker) | PHQ-9 | Physical activity correlates with reduction in the level of depression and anxiety. |
| Saeb et al [69] | 48 healthy subjects | GPS | PHQ-9 | GPS correlates with depression differently on weekdays and weekends. |
| Place et al [70] | 73 subjects with at least one symptom of depression | GPS, accelerometer, gyroscope, phone calls, messages, microphone, and screen on/off | Semi-structured clinical interview | Physical activity and phone usage relate to depression symptoms. |
| Saeb et all [71] | 206 healthy subjects | GPS, accelerometer, gyroscope (Android activity-recognition APIp), light sensor, microphone, screen on/off, phone calls, and messages | PHQ-9 and GAD-7q | No consistent relationship between GPS-based semantic location and depression or anxiety. |
| Faurholt-Jepsen et al [72] | 61 subjects with bipolar disorders | Phone calls and messages | HAMD and YMRS | Significant correlation between depressive and manic symptoms and phone usage. |
| Sabatelli et al [73] | 7 subjects with bipolar disorders | Wi-Fi-based position | HAMD and YMRS | Weak negative correlation between staying in clinics and self-reported state. |
| Rabbi et al [74] | 8 healthy subjects (elders) | Accelerometer, gyroscope, barometer, and microphone | Friendship Scale, SF-36r, CES-Ds, and YPASt | No clear relationship between sensor data and administered assessment scales. |
| Doryab et al [75] | 3 healthy subjects | GPS, accelerometer, gyroscope, microphone, and light sensor | CES-D | Correlation between depression scales and sensor data. |
| Farhan et al [76] | 60 healthy subjects | GPS, accelerometer, gyroscope, microphone, phone lock and unlock, light sensor, and phone call duration | PHQ-9 | Correlation between PHQ-9 scores and all the sensor data is pointed out. |
| Canzian et al [77] | 28 healthy subjects | GPS | PHQ-8 | Significant correlation between mobility patterns and depressive mood. |
| Zulueta et al [78] | 16 subjects with bipolar disorders | Phone keyboard usage | HAMD and YMRS | Accelerometer activity while typing, number of exchanged messages, and typing errors correlate with depression and mania scores. |
| Sano et al [79] | 201 healthy subjects | Skin conductance, skin temperature, accelerometer, ambient light, GPS, phone calls, messages, app usage, and phone lock and unlock | PSS and MCSu | Skin conductance relates to MCS, skin temperature, and phone usage timing and duration; GPS relates both to PSS and MCS. |
| Tron et al [80] | 25 subjects with schizophrenia, DSM-IV | Accelerometer, light, temperature | PANSS | Physical activity relates to PANSS. |
| Cella et al [81] | 30 subjects with schizophrenia, DSM-IV, and 25 controls | Accelerometer, skin conductance, heart rate variability, and interbeat intervals | PANSS | Interbeat intervals negatively correlate with positive symptoms; movement negatively correlates with negative symptoms. |
aPHQ-9: Patient Health Questionnaire-9.
bPSS: Perceived Stress Scale.
cUCLA: University of California, Los Angeles.
dHAMD: Hamilton Depression Rating Scale.
eYMRS: Young Mania Rating Scale.
fSIAS: Social Interaction Anxiety Scale.
gDASS-21: Depression, Anxiety, and Stress Scale.
hPSQI: Pittsburgh Sleep Quality Index.
iMEQ: Horne-Ostberg Morningness-Eveningness Questionnaire.
jMCS for mental health: Short Form-12 Physical and Mental Health Composite Scale.
kPHQ-2: Patient Health Questionnaire-2.
lPHQ-8: Patient Health Questionnaire-8.
mQID: Quick Inventory of Depression.
nDSM-IV: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition.
oPANSS: Positive and Negative Syndrome Scale.
pAPI: application programming interface.
qGAD-7: General Anxiety Disorder questionnaire.
rSF-36: Short Form-36 Health Survey.
sCES-D: Center for Epidemiologic Studies-Depression scale.
tYPAS: Yale Physical Activity Survey.
uMCS: Mental Component Summary.