Digital phenotyping in psychiatry: A scoping review

Abstract

BACKGROUND:

Digital phenotyping has been defined as the moment-by-moment assessment of an illness state through digital means, promising objective, quantifiable data on psychiatric patients’ conditions, and could potentially improve diagnosis and management of mental illness. As it is a rapidly growing field, it is to be expected that new literature is being published frequently.

OBJECTIVE:

We conducted this scoping review to assess the current state of literature on digital phenotyping and offer some discussion on the current trends and future direction of this area of research.

METHODS:

We searched four databases, PubMed, Ovid MEDLINE, PsycINFO and Web of Science, from inception to August 25 ${}^{\text{th}}$ , 2021. We included studies written in English that 1) investigated or applied their findings to diagnose psychiatric disorders and 2) utilized passive sensing for management or diagnosis. Protocols were excluded. A narrative synthesis approach was used, due to the heterogeneity and variability in outcomes and outcome types reported.

RESULTS:

Of 10506 unique records identified, we included a total of 107 articles. The number of published studies has increased over tenfold from 2 in 2014 to 28 in 2020, illustrating the field’s rapid growth. However, a significant proportion of these (49% of all studies and 87% of primary studies) were proof of concept, pilot or correlational studies examining digital phenotyping’s potential. Most (62%) of the primary studies published evaluated individuals with depression (21%), BD (18%) and SZ (23%) (Appendix 1).

CONCLUSION:

There is promise shown in certain domains of data and their clinical relevance, which have yet to be fully elucidated. A consensus has yet to be reached on the best methods of data collection and processing, and more multidisciplinary collaboration between physicians and other fields is needed to unlock the full potential of digital phenotyping and allow for statistically powerful clinical trials to prove clinical utility.

Keywords

Digital phenotyping digital sensing psychiatry scoping review artificial intelligence machine learning mobile health

1. Introduction

Digital phenotyping refers to the moment-by-moment assessment of an illness state through digital means, by means of mobile devices and their accompanying applications and wearable sensors [1, 2, 3]. Such assessment is of importance, as it could be used to measure human’s behavior and this information will be essential for clinical assessment; such information could be used in the prediction of the clinical status of individuals and could help to facilitate the delivery of on-demand interventions. The data that is collated and utilized for the purposes of digital phenotyping could be classified into two categories, that of active and passive data. “Active data” refers to data that requires individuals’ action or inputs, whereas “passive data” refers to data that is collated without individuals’ action or inputs, and usually by means of sensors like the global positioning system (GPS), other biosensors such as heart rate monitoring, and event data like patients’ call logs or their social media usage [4, 5, 6].

With the penetrance of smartphones ever-increasing worldwide (up from 49.35% in 2016 to 78.05% in 2020) [7], it is expected that more individuals will be using smartphones; and with the advances in smartphone technologies, more data could thus be collated. Digital phenotyping has been well explored in other fields of medicine. Data collated by means of digital phenotyping has been used to quantify ALS progression [8, 9] and even assess mobility and quality of life in spine disease patients [10]. In specialties like psychiatry, there have already been studies reporting the findings using digital phenotyping in various psychiatric conditions. Some examples include mobile phone detection of semantic locations visited and its relationship to depression and anxiety [11], predicting mood disturbance severity with mobile phone keystroke metadata in patients with Bipolar Disorder (BD) [12], prediction of relapse risk in Schizophrenia (SZ) through smartphone derived estimates of mobility, sociability, screen time and sleep [13] or even detecting recovery problems through linguistic analysis in an online substance abuse forum [14]. This is a exciting and fast-growing field in psychiatry, with promise to revolutionize practice with a shift toward “precision psychiatry” – more personalized, targeted treatment with more objective data on patients’ symptoms [15]. The utilization of passive data parameters also reduces common confounders in psychiatric research and evaluation, that of patient recall bias and the need for self-reporting. In general, studies have also reported that patients are mostly willing to use digital phenotyping and share the data with their healthcare providers if privacy and data security is assured [16, 17, 18].

There are, to our knowledge, currently no reviews that have broadly assessed the state of digital phenotyping research in psychiatry. As a rapidly growing field, it is to be expected that new literature is being published frequently, and thus we conducted this scoping review to assess the current state of literature on digital phenotyping and offer some discussion on the current trends and future direction of this area of research.

2. Methods

2.1 Information sources and search strategy

A comprehensive literature search in bibliographic databases was conducted, spanning PubMed, Ovid MEDLINE, PsycINFO and Web of Science. The search strategy was crafted in consultation with a librarian at the Lee Kong Chian School of Medicine, Nanyang Technological University Singapore. The target population was individuals with psychiatric disorders, and the intervention used was that of digital phenotyping. The search strategy was developed on PubMed, and subsequently adapted to the remaining search engines. MeSH headings were used where available, and related terms were included in the search strategy as well (the complete search strategy is listed in Appendix 2). All searches were conducted from the database inception through to August 25, 2021.

2.2 Inclusion and exclusion criteria

Reviewers independently screened articles for eligibility and obtained full texts of all potentially relevant papers. We included studies written in English that 1) investigated or applied their findings to diagnosed Psychiatric disorders and 2) utilized digital phenotyping for management or diagnosis. Protocols were excluded.

2.3 Data extraction

Search results were exported into EndNote X9 and subsequently uploaded into Covidence, where duplicates were removed, screening conducted, and data extracted.

The following was extracted from included studies: 1) Publication information (author, year of publication), 2) the country in which study was conducted, 3) Study design, 4) Population description (cohort sizes, Psychiatric diagnoses of interest), 5) Types of sensing used (active/passive) and data collected and 6) Outcomes.

2.4 Analysis

A narrative synthesis approach was used, and processing of extracted data and creation of figures were handled in Excel.

Figure 1.

PRISMA flowchart of the study selection process.

3. Results

As shown in Fig. 1, there were an initial 10506 records after removal of 9082 duplicates. Titles and abstracts were screened to determine eligibility, and 150 full-text records were obtained and assessed against the inclusion and exclusion criteria. A final 107 were included. The number of published studies per year increased over tenfold from 2 in 2014 to 28 in 2020, illustrating the field’s rapid growth (graph available in Appendix 3).

3.1 Study characteristics

Of the included studies, a significant proportion (49% of all studies and 87% of primary studies) were proof of concept, pilot or correlational studies examining digital phenotyping’s potential (Table 1). As such, they typically had smaller cohort sizes, with a mean of 60.2 and median 50. These highlight the fact that the field is still developing and attempting to determine the feasibility of various methods and interventions. Thus, studies validating methods of digital phenotyping and clinical applications are relatively fewer [19, 20].

Table 1
Number of studies included and type of study

Study type	$N=$
Systematic review	7
Randomized controlled trial	1
Cross-sectional study	1
Proof of concept/pilot study	39
Correlational study	13
Qualitative research	4
Narrative reviews (including texts and opinions)	38
Case series/reports	2
Others (scoping review, guideline proposal)	2
Total	107

3.2 Geographical distribution of studies

When analyzing the country in which studies were conducted, we found that the majority ( $n=$ 120) were conducted in developed economies – particularly the United States, which was involved in 65 papers, while only 6 studies were conducted in developing economies and 8 in Asian countries as defined by the U.N. [21] (graphical representation available in Appendix 4).

3.3 Distribution of research among various psychiatric conditions

It appears that most (62%) of the primary studies published evaluated individuals with depression (21%), BD (18%) and SZ (23%).

3.4 Types of data collected

Most correlational or proof-of-concept studies included used a combination of both active and passive sensing ( $n=$ 30), with the former typically being EMAs, as compared to studies which used passive sensing alone ( $n=$ 21). Active data was most used in correlational studies, providing a means to administer clinically validated inventories with greater temporal resolution to better correlate passively sensed data with the participant’s current state.

Most studies ( $n=$ 39) utilized smartphone-only sensing, while a minority ( $n=$ 9) used a combination of smartphone and smartwatches/wristbands. Remaining studies used other sensors ( $n=$ 1), only wristbands ( $n=$ 1) or were web-based ( $n=$ 5).

Parameters measured varied greatly, including but not limited to geolocation, accelerometer (and derived contextual cues), call/text/communication app logs, microphone audio, video, ambient light, heart rate, electrodermal activity, screen status and app usage

4. Discussion

The top three studied disorders were SZ, BD, and Depression as shown in Fig. 2. According to Rehm et al. [75], these diseases constitute an estimated 41% of global mental health disease burden as measured in disability-adjusted life years (DALYs), and it is expected that these conditions are highly evaluated. There are a considerable number of narrative reviews ( $n=$ 14) or opinion pieces ( $n=$ 24) currently published for digital phenotyping in various subspecialties of psychiatry. This is to be expected since the field is still in its infancy, and with the heterogeneous nature of outcomes there is much debate on the most appropriate mechanisms of data collection and processing to achieve predictive accuracy.

Figure 2.

Analysis of studies by psychiatric condition.

Our first key finding is a lack of consensus on which data streams are relevant, and in what resolution and for which psychiatric conditions. Most primary studies (30/51) utilized a combination of passive sensing and active sensing, typically Ecological Momentary Assessments (EMAs). This was usually to deliver standardized clinical measures such as the PHQ-9 or GAD-7 in comparable spatiotemporal resolution to assess the clinical utility of passively sensed data more appropriately. As described by Cohen et al. [76, 77, 78], validity against “gold standard” measures is problematic if digital phenotyping and “gold standard” criterion are not matched in time and space. Thus, considering spatiotemporal resolution is of utmost importance when validating correlations and drawing conclusions. There is great diversity on which data is collected, shown in the non-exhaustive list in Section 3.5. The most gathered data was geolocation ( $n=$ 20), call/text logs ( $n=$ 19), and screen status/app usage ( $n=$ 17). We postulate that this is because data streams are easily accessible and more ‘established’, beginning to be validated in different contexts and populations. This is supported by Bai et al. [36], who found that, in machine learning models, “the best combination of data was call logs, sleep data, step count data and HR data reaching 80% accuracy”. The time frame of studies we found also varied greatly, with study duration ranging from weeks up to years. There also appears to be some segregation between studies that focus on short- and long-term predictions. For example, there are several studies regarding real-time prediction of suicide risk using digital phenotyping [48, 79, 80, 81]; the increased interest in this niche area could be because situations regarding suicide require quick identification and intervention. On the other hand, there are studies spanning years, such as the work by Jacobson et al., in predicting deterioration in anxiety disorder symptoms across 17–18 years. [37] Such models which predict long term disease course could allow better risk stratification and intervention planning, thereby improving care of patients.

Secondly, there is also a lack of consensus on the best ways to approach data processing, with great diversity on what data is collected, as listed in the non-exhaustive list in Section 3.5. Each of these data streams need to be correlated individually as well as in combination with other collected data, thus presenting the gargantuan problem of data processing. Fortunately, with the advancement of machine learning, there is hope for less labor-intensive yet more efficacious ways of identifying clinically relevant patterns in data collected. A recent review by Antosik-Wojcinska et al. [82] found that “State-of-the-art machine learning solutions have approximately accuracy ranging from 67% to 97%”, proving the feasibility of using machine learning. This is especially relevant when considering that some studies show that correlations between collected data and standardized clinical measures vary significantly on an individual level – weakening or even causing non-significance at cohort level, [37, 67, 68, 73] thus contributing to the heterogeneity of results. Thus, a shift toward individualized methods of data analysis and validation of passively sensed data should be explored, as demonstrated by Wisniewski et al. [73]. Machine learning can be leveraged upon to train individual-specific models, possibly conferring further improvements in the clinical utility of digital phenotyping.

Although this study was intended to be a scoping review, PRISMA guidelines [84] were followed, with the exception of quality assessments. Our search, strategy was also broad and comprehensive. However, as this review was meant to be a broad overview of the state of literature, we did not elaborate excessively on current findings in each type of data or psychiatric illness due to the large variety of articles included. The project time frame was also limited to 6 weeks, and we did not hand-search references of identified articles to be included to ensure that all relevant studies were captured. We also did not search other databases such as CINAHL and Embase.

There are several research and clinical implications arising from this study. With regards to research implications, there seems to be a consensus forming on which data are more relevant than others, and which data processing techniques are promising. We must consider validating these established relationships with more statistically powerful methods. However, due to possible poor generalizability and high inter-individual variance [11], there is a need for RCTs that are specific and targeted in scope. Further research should investigate the feasibility of creating individual correlation models to maximize monitoring accuracy and thus clinical utility. Multidisciplinary collaboration between physicians and other fields, such as software engineers and computer scientists, must continue to refine methods of analysis and gain deeper understandings of gathered data to apply them to the clinical context [83]. We believe that with better subcategorization of studies and more data on specific correlations in defined spatiotemporal contexts, the field can begin to move toward validating said correlations with standardized measures and thus improve the clinical utility of digital phenotyping.

5. Conclusion

Most primary studies show promise in certain domains and relationships, which have yet to be fully elucidated, requiring more investigation into sub-categories and reasons behind this.

Data availability

All available data have been included in the manuscript.

Author contributions

AC & MZ jointly conceptualized the study. AC and MZ reviewed the literature and obtained the primary data. AC performed the analysis with guidance from MZ. MZ verified the data-analysis. AC wrote the first draft of the manuscript, which MZ provided critical input. All authors read and approved of the manuscript prior to submission.

Funding

MWZ is supported by a grant under the Singapore Ministry of Health’s National Medical Research Council (grant number NMRC/Fellowship/0048/2017) for PhD training. The funding source was not involved in any part of this project.

Supplementary data

The appendices are available from https://dx-doi-org-s.web.bisu.edu.cn/10.3233/THC-213648.

Footnotes

Conflict of interest

None of the authors have any competing interests.

References

Jain

Powers

Hawkins

Brownstein

. The digital phenotype. Nature Biotechnology. 2015; 33(5): 462-3. doi: 10.1038/nbt.3223.

Insel

. Digital Phenotyping: Technology for a New Science of Behavior. Jama. 2017 Oct 3; 318(13): 1215-6. PMID: 28973224. doi: 10.1001/jama.2017.11295.

Dagum

. Digital biomarkers of cognitive function. NPJ Digital Medicine. 2018; 1(1): 10. doi: 10.1038/s41746-018-0018-4.

Torous

Friedman

Keshavan

. Smartphone Ownership and Interest in Mobile Applications to Monitor Symptoms of Mental Health Conditions. Jmir Mhealth and Uhealth. 2014 Jan-Mar; 2(1). PMID: WOS:000209894900012. doi: 10.2196/mhealth.2994.

Derks

Klaassen

Westerhof

Bohlmeijer

Noordzij

. Development of an Ambulatory Biofeedback App to Enhance Emotional Awareness in Patients with Borderline Personality Disorder: Multicycle Usability Testing Study. JMIR Mhealth Uhealth. 2019 Oct 15; 7(10): e13479. PMID: 31617851. doi: 10.2196/13479.

Daus

Kislicyn

Heuer

Backenstrass

. Disease management apps and technical assistance systems for bipolar disorder: Investigating the patients’ point of view. J Affect Disord. 2018 Mar 15; 229: 351-7. PMID: 29331693. doi: 10.1016/j.jad.2017.12.059.

Statista. Global smartphone penetration rate as share of population from 2016 to 2020. 2021.

Connaghan

Green

Paganoni

Chan

Weber

Collins

, et al. Use of Beiwe Smartphone App to Identify and Track Speech Decline in Amyotrophic Lateral Sclerosis (ALS). 2019; pp. 4504-8.

Berry

Paganoni

Carlson

Burke

Weber

Staples

, et al. Design and results of a smartphone-based digital phenotyping study to quantify ALS progression. Annals of Clinical and Translational Neurology. 2019; 6(5): 873-81. doi: 10.1002/acn3.770.

10.

Cote

Barnett

Onnela

J-P

Smith

. Digital Phenotyping in Patients with Spine Disease: A Novel Approach to Quantifying Mobility and Quality of Life. World Neurosurgery. 2019; 126: e241-e9. doi: 10.1016/j.wneu.2019.01.297.

11.

Saeb

Lattie

Kording

Mohr

. Mobile Phone Detection of Semantic Location and Its Relationship to Depression and Anxiety. Jmir Mhealth and Uhealth. 2017 Aug; 5(8). PMID: WOS:000410033700016. doi: 10.2196/mhealth.7297.

12.

Zulueta

Piscitello

Rasic

Easter

Babu

Langenecker

, et al. Predicting Mood Disturbance Severity with Mobile Phone Keystroke Metadata: A BiAffect Digital Phenotyping Study. J Med Internet Res. 2018 Jul 20; 20(7): e241. PMID: 30030209. doi: 10.2196/jmir.9775.

13.

Henson

D’Mello

Vaidyam

Keshavan

Torous

. Anomaly detection to predict relapse risk in schizophrenia. Transl Psychiatry. 2021 Jan 11; 11(1): 28. PMID: 33431818. doi: 10.1038/s41398-020-01123-7.

14.

Kornfield

Sarma

Shah

McTavish

Landucci

Pe-Romashko

, et al. Detecting Recovery Problems Just in Time: Application of Automated Linguistic Analysis and Supervised Machine Learning to an Online Substance Abuse Forum. J Med Internet Res. 2018 Jun 12; 20(6): e10136. PMID: 29895517. doi: 10.2196/10136.

15.

Fernandes

Williams

Steiner

Leboyer

Carvalho

Berk

. The new field of ‘precision psychiatry’. BMC Med. 2017 Apr 13; 15(1): 80. PMID: 28403846. doi: 10.1186/s12916-017-0849-x.

16.

Nicholas

Shilton

Schueller

Gray

Kwasny

Mohr

. The Role of Data Type and Recipient in Individuals’ Perspectives on Sharing Passively Collected Smartphone Data for Mental Health: Cross-Sectional Questionnaire Study. JMIR Mhealth Uhealth. 2019 Apr 5; 7(4): e12578. PMID: 30950799. doi: 10.2196/12578.

17.

Torous

Chan

Tan

SYM

Behrens

Mathew

Conrad

, et al. Patient Smartphone Ownership and Interest in Mobile Apps to Monitor Symptoms of Mental Health Conditions: A Survey in Four Geographically Distinct Psychiatric Clinics. Jmir Mental Health. 2014 Jul-Dec; 1(1). PMID: WOS: 000414977300003. doi: 10.2196/mental.4004.

18.

Proudfoot

Parker

Pavlovic

Manicavasagar

Adler

Whitton

. Community Attitudes to the Appropriation of Mobile Phones for Monitoring and Managing Depression, Anxiety, and Stress. Journal of Medical Internet Research. 2010; 12(5). PMID: WOS:000285714300011. doi: 10.2196/jmir.1475.

19.

Seppala

De Vita

Jamsa

Miettunen

Isohanni

Rubinstein

, et al. Mobile Phone and Wearable Sensor-Based mHealth Approach for Psychiatric Disorders and Symptoms: Systematic Review and Link to the m-RESIST Project. Jmir Mental Health. 2019 Feb; 6(2). PMID: WOS:000460606700001. doi: 10.2196/mental.9819.

20.

Faurholt-Jepsen

Bauer

Kessing

. Smartphone-based objective monitoring in bipolar disorder: status and considerations. International Journal of Bipolar Disorders. 2018 Jan; 6. PMID: WOS: 000423219800001. doi: 10.1186/s40345-017-0110-8.

21.

Nations

. World Economic Situation and Prospects 2020. 2020.

22.

Van Til

McInnis

Cochran

. A comparative study of engagement in mobile and wearable health monitoring for bipolar disorder. Bipolar Disord. 2020 Mar; 22(2): 182-90. PMID: 31610074. doi: 10.1111/bdi.12849.

23.

Stanislaus

Vinberg

Melbye

Frost

Busk

Bardram

, et al. Daily self-reported and automatically generated smartphone-based sleep measurements in patients with newly diagnosed bipolar disorder, unaffected first-degree relatives and healthy control individuals. Evidence-Based Mental Health. 2020 Nov; 23(4): 146-53. PMID: WOS: 000584955400006. doi: 10.1136/ebmental-2020-300148.

24.

Jacobson

Summers

Wilhelm

. Digital Biomarkers of Social Anxiety Severity: Digital Phenotyping Using Passive Smartphone Sensors. J Med Internet Res. 2020 May 29; 22(5): e16875. PMID: 32348284. doi: 10.2196/16875.

25.

Stanislaus

Vinberg

Melbye

Frost

Busk

Bardram

, et al. Smartphone-based activity measurements in patients with newly diagnosed bipolar disorder, unaffected relatives and control individuals. International Journal of Bipolar Disorders. 2020 Nov; 8(1). PMID: WOS: 000588316000001. doi: 10.1186/s40345-020-00195-0.

26.

Ryan

Babu

Easter

Saunders

Lee

Klasnja

, et al. A Smartphone App to Monitor Mood Symptoms in Bipolar Disorder: Development and Usability Study. Jmir Mental Health. 2020 Sep; 7(9). PMID: WOS:000571663200001. doi: 10.2196/19476.

27.

Ebner-Priemer

Muhlbauer

Neubauer

Hill

Beier

Santangelo

, et al. Digital phenotyping: towards replicable findings with comprehensive assessments and integrative models in bipolar disorders. International Journal of Bipolar Disorders. 2020 Dec; 8(1). PMID: WOS:000590987100001. doi: 10.1186/s40345-020-00210-4.

28.

Daus

Blocher

Egeler

De Klerk

Stork

Backenstrass

. Development of an Emotion-Sensitive mHealth Approach for Mood-State Recognition in Bipolar Disorder. Jmir Mental Health. 2020 Jul; 7(7). PMID: WOS:000545285100001. doi: 10.2196/14267.

29.

Ranjan

Rashid

Stewart

Conde

Begale

Verbeeck

, et al. RADAR-Base: Open Source Mobile Health Platform for Collecting, Monitoring, and Analyzing Data Using Sensors, Wearables, and Mobile Devices. Jmir Mhealth and Uhealth. 2019 Aug; 7(8). PMID: WOS:000481504800001. doi: 10.2196/11734.

30.

Jagesar

Roozen

van der Heijden

Ikani

Tyborowska

Penninx

, et al. Digital phenotyping and the COVID-19 pandemic: Capturing behavioral change in patients with psychiatric disorders. Eur Neuropsychopharmacol. 2021 Jan; 42: 115-20. PMID: 33298386. doi: 10.1016/j.euroneuro.2020.11.012.

31.

Gloster

Meyer

Klotsche

Villanueva

Block

Benoy

, et al. The spatiotemporal movement of patients in and out of a psychiatric hospital: an observational GPS study. Bmc Psychiatry. 2021 Mar; 21(1). PMID: WOS:000634792800002. doi: 10.1186/s12888-021-03147-9.

32.

Wahle

Kowatsch

Fleisch

Rufer

Weidt

. Mobile Sensing and Support for People With Depression: A Pilot Trial in the Wild. Jmir Mhealth and Uhealth. 2016 Jul-Sep; 4(3). PMID: WOS:000391887800013. doi: 10.2196/mhealth.5960.

33.

Sverdlov

Curcic

Hannesdottir

Gou

De Luca

Ambrosetti

, et al. A Study of Novel Exploratory Tools, Digital Technologies, and Central Nervous System Biomarkers to Characterize Unipolar Depression. Frontiers in Psychiatry. 2021 May; 12. PMID: WOS:000652282700001. doi: 10.3389/fpsyt.2021.640741.

34.

Scherr

Goering

. Is a Self-Monitoring App for Depression a Good Place for Additional Mental Health Information? Ecological Momentary Assessment of Mental Help Information Seeking among Smartphone Users. Health Commun. 2020 Jul; 35(8): 1004-12. PMID: 31025888. doi: 10.1080/10410236.2019.1606135.

35.

Jacobson

Chung

. Passive Sensing of Prediction of Moment-To-Moment Depressed Mood among Undergraduates with Clinical Levels of Depression Sample Using Smartphones. Sensors (Basel). 2020 Jun 24; 20(12). PMID: 32599801. doi: 10.3390/s20123572.

36.

Bai

Xiao

Guo

Zhu

Wang

, et al. Tracking and Monitoring Mood Stability of Patients With Major Depressive Disorder by Machine Learning Models Using Passive Digital Data: Prospective Naturalistic Multicenter Study. JMIR Mhealth Uhealth. 2021 Mar 8; 9(3): e24365. PMID: 33683207. doi: 10.2196/24365.

37.

Jacobson

Lekkas

Huang

Thomas

. Deep learning paired with wearable passive sensing data predicts deterioration in anxiety disorder symptoms across 17–18 years. J Affect Disord. 2021 Mar 1; 282: 104-11. PMID: 33401123. doi: 10.1016/j.jad.2020.12.086.

38.

Tomczak

Wójcikowski

Listewnik

Pankiewicz

Majchrowicz

Jȩdrzejewska-Szczerska

. Support for Employees with ASD in the Workplace Using a Bluetooth Skin Resistance Sensor – A Preliminary Study. Sensors (Basel). 2018 Oct 19; 18(10). PMID: 30347649. doi: 10.3390/s18103530.

39.

Palmius

Tsanas

Saunders

KEA

Bilderbeck

Geddes

Goodwin

, et al. Detecting Bipolar Depression From Geographic Location Data. IEEE Trans Biomed Eng. 2017 Aug; 64(8): 1761-71. PMID: 28113247. doi: 10.1109/tbme.2016.2611862.

40.

Raugh

James

Gonzalez

Chapman

Cohen

Kirkpatrick

, et al. Geolocation as a Digital Phenotyping Measure of Negative Symptoms and Functional Outcome. Schizophrenia Bulletin. 2020 Nov; 46(6): 1596-607. PMID: WOS:000607041400030. doi: 10.1093/schbul/sbaa121.

41.

Tariq

Akhtar

Afzal

Khalid

Mufti

Hussain

, et al. A Novel Co-Training-Based Approach for the Classification of Mental Illnesses Using Social Media Posts. Ieee Access. 2019; 7: 166165-72. PMID: WOS:000498717700001. doi: 10.1109/access.2019.2953087.

42.

Cohen

Fedechko

Schwartz

Foltz

Bernstein

, et al. Ambulatory Vocal Acoustics, Temporal Dynamics, and Serious Mental Illness. Journal of Abnormal Psychology. 2019 Feb; 128(2): 97-105. PMID: WOS:000457471600001. doi: 10.1037/abn0000397.

43.

Albin-Rodriguez

Ricoy-Cano

de-la-Fuente-Robles

Espinilla-Estevez

. Fuzzy Protoform for Hyperactive Behaviour Detection Based on Commercial Devices. International Journal of Environmental Research and Public Health. 2020 Sep; 17(18). PMID: WOS:000580300000001. doi: 10.3390/ijerph17186752.

44.

Lam

Wang

JKY

Han

Zheng

Kam

CHC

, et al. SmartMood: Toward Pervasive Mood Tracking and Analysis for Manic Episode Detection. Ieee Transactions on Human-Machine Systems. 2015 Feb; 45(1): 126-31. PMID: WOS:000348072800012. doi: 10.1109/thms.2014.2360469.

45.

Lekkas

Jacobson

. Using artificial intelligence and longitudinal location data to differentiate persons who develop posttraumatic stress disorder following childhood trauma. Scientific Reports. 2021 May; 11(1). PMID: WOS:000656952400060. doi: 10.1038/s41598-021-89768-2.

46.

Wang

Vouk

Heaukulani

Buddhika

Martanto

Lee

, et al. HOPES: An Integrative Digital Phenotyping Platform for Data Collection, Monitoring, and Machine Learning. J Med Internet Res. 2021 Mar 15; 23(3): e23984. PMID: 33720028. doi: 10.2196/23984.

47.

Lee

Rho

Yook

Park

Jang

Park

, et al. Design, Development and Implementation of a Smartphone Overdependence Management System for the Self-Control of Smart Devices. Applied Sciences-Basel. 2016 Dec; 6(12). PMID: WOS:000391626800015. doi: 10.3390/app6120440.

48.

Glenn

Nobles

Barnes

Teachman

. Can Text Messages Identify Suicide Risk in Real Time? A Within-Subjects Pilot Examination of Temporally Sensitive Markers of Suicide Risk. Clinical Psychological Science. 2020 Jul; 8(4): 704-22. PMID: WOS:000536549700001. doi: 10.1177/2167702620906146.

49.

Staples

Torous

Barnett

Carlson

Sandoval

Keshavan

, et al. A comparison of passive and active estimates of sleep in a cohort with schizophrenia. Npj Schizophrenia. 2017 Oct; 3. PMID: WOS:000414578100001. doi: 10.1038/s41537-017-0038-0.

50.

Reinertsen

Osipov

Liu

Kane

Petrides

Clifford

. Continuous assessment of schizophrenia using heart rate and accelerometer data. Physiological Measurement. 2017 Jul; 38(7): 1456-71. PMID: WOS:000404541400005. doi: 10.1088/1361-6579/aa724d.

51.

Raugh

James

Gonzalez

Chapman

Cohen

Kirkpatrick

, et al. Digital phenotyping adherence, feasibility, and tolerability in outpatients with schizophrenia. J Psychiatr Res. 2021 Jun; 138: 436-43. PMID: 33964681. doi: 10.1016/j.jpsychires.2021.04.022.

52.

He-Yueya

Buck

Campbell

Choudhury

Kane

Ben-Zeev

, et al. Assessing the relationship between routine and schizophrenia symptoms with passively sensed measures of behavioral stability. Npj Schizophrenia. 2020 Nov; 6(1). PMID: WOS:000595712100001. doi: 10.1038/s41537-020-00123-2.

53.

Cohen

Cowan

Schwartz

Kirkpatrick

Raugh

, et al. Ambulatory digital phenotyping of blunted affect and alogia using objective facial and vocal analysis: Proof of concept. Schizophr Res. 2020 Jun; 220: 141-6. PMID: 32247747. doi: 10.1016/j.schres.2020.03.043.

54.

Ben-Zeev

Wang

Abdullah

Brian

Scherer

Mistler

, et al. Mobile Behavioral Sensing for Outpatients and Inpatients With Schizophrenia. Psychiatr Serv. 2016 May 1; 67(5): 558-61. PMID: 26695497. doi: 10.1176/appi.ps.201500130.

55.

Barnett

Torous

Staples

Sandoval

Keshavan

Onnela

. Relapse prediction in schizophrenia through digital phenotyping: a pilot study. Neuropsychopharmacology. 2018 Jul; 43(8): 1660-6. PMID: 29511333. doi: 10.1038/s41386-018-0030-z.

56.

Leonard

Silverman

Sherpa

Naegle

Kim

Coffman

, et al. Mobile Health Technology Using a Wearable Sensorband for Female College Students With Problem Drinking: An Acceptability and Feasibility Study. Jmir Mhealth and Uhealth. 2017 Jul; 5(7). PMID: WOS:000431849300001. doi: 10.2196/mhealth.7399.

57.

Kornfield

Toma

Shah

Moon

Gustafson

. What Do You Say Before You Relapse? How Language Use in a Peer-to-peer Online Discussion Forum Predicts Risky Drinking among Those in Recovery. Health Commun. 2018 Sep; 33(9): 1184-93. PMID: 28792228. doi: 10.1080/10410236.2017.1350906.

58.

Bae

Chung

Ferreira

Dey

Suffoletto

. Mobile phone sensors and supervised machine learning to identify alcohol use events in young adults: Implications for just-in-time adaptive interventions. Addict Behav. 2018 Aug; 83: 42-7. PMID: 29217132. doi: 10.1016/j.addbeh.2017.11.039.

59.

Rosenthal

Zhou

Booth

. Association between mobile phone screen time and depressive symptoms among college students: A threshold effect. Human Behavior and Emerging Technologies. 2021 Jul; 3(3): 432-40. PMID: WOS:000674324700009. doi: 10.1002/hbe2.256.

60.

Saeb

Zhang

Karr

Schueller

Corden

Kording

, et al. Mobile Phone Sensor Correlates of Depressive Symptom Severity in Daily-Life Behavior: An Exploratory Study. Journal of Medical Internet Research. 2015 Jul; 17(7). PMID: WOS:000358010500002. doi: 10.2196/jmir.4273.

61.

Saeb

Lattie

Schueller

Kording

Mohr

. The relationship between mobile phone location sensor data and depressive symptom severity. Peerj. 2016 Sep; 4. PMID: WOS:000385572500009. doi: 10.7717/peerj.2537.

62.

Gillett

McGowan

Palmius

Bilderbeck

Goodwin

Saunders

KEA

. Digital Communication Biomarkers of Mood and Diagnosis in Borderline Personality Disorder, Bipolar Disorder, and Healthy Control Populations. Frontiers in Psychiatry. 2021 Apr; 12. PMID: WOS:000642092000001. doi: 10.3389/fpsyt.2021.610457.

63.

Sarda

Munuswamy

Sarda

Subramanian

. Using Passive Smartphone Sensing for Improved Risk Stratification of Patients With Depression and Diabetes: Cross-Sectional Observational Study. Jmir Mhealth and Uhealth. 2019 Jan; 7(1). PMID: WOS:000457486600001s. doi: 10.2196/11041.

64.

Dáu

Callinan

Mayes

Smith

. Postpartum depressive symptoms and maternal sensitivity: an exploration of possible social media-based measures. Arch Womens Ment Health. 2017 Feb; 20(1): 221-4. PMID: 27416930. doi: 10.1007/s00737-016-0650-4.

65.

Chow

Fua

Huang

Bonelli

Xiong

Barnes

, et al. Using Mobile Sensing to Test Clinical Models of Depression, Social Anxiety, State Affect, and Social Isolation Among College Students. J Med Internet Res. 2017 Mar 3; 19(3): e62. PMID: 28258049. doi: 10.2196/jmir.6820.

66.

Schlier

Krkovic

Clamor

Lincoln

. Autonomic arousal during psychosis spectrum experiences: Results from a high resolution ambulatory assessment study over the course of symptom on- and offset. Schizophr Res. 2019 Oct; 212: 163-70. PMID: 31422861. doi: 10.1016/j.schres.2019.07.046.

67.

Henson

Rodriguez-Villa

Torous

. Investigating Associations Between Screen Time and Symptomatology in Individuals With Serious Mental Illness: Longitudinal Observational Study. Journal of Medical Internet Research. 2021 Mar; 23(3). PMID: WOS:000636163000003. doi: 10.2196/23144.

68.

Henson

Barnett

Keshavan

Torous

. Towards clinically actionable digital phenotyping targets in schizophrenia. Npj Schizophrenia. 2020 May; 6(1). PMID: WOS:000530531200001. doi: 10.1038/s41537-020-0100-1.

69.

Fulford

Mote

Gonzalez

Abplanalp

Zhang

Luckenbaugh

, et al. Smartphone sensing of social interactions in people with and without schizophrenia. J Psychiatr Res. 2021 May; 137: 613-20. PMID: 33190842. doi: 10.1016/j.jpsychires.2020.11.002.

70.

Lauckner

Desrosiers

Muilenburg

Killanin

Genter

Kershaw

. Social media photos of substance use and their relationship to attitudes and behaviors among ethnic and racial minority emerging adult men living in low-income areas. Journal of Adolescence. 2019 Dec; 77: 152-62. PMID: WOS:000501942900016. doi: 10.1016/j.adolescence.2019.10.013.

71.

Martinez-Martin

Greely

Cho

. Ethical Development of Digital Phenotyping Tools for Mental Health Applications: Delphi Study. JMIR MHealth and UHealth. 2021; 9(7): e27343. PMID: 34319252.

72.

Cohen

Fedechko

Schwartz

Foltz

Bernstein

, et al. Psychiatric Risk Assessment from the Clinician’s Perspective: Lessons for the Future. Community Ment Health J. 2019 Oct; 55(7): 1165-72. PMID: 31154587. doi: 10.1007/s10597-019-00411-x.

73.

Wisniewski

Henson

Torous

. Using a Smartphone App to Identify Clinically Relevant Behavior Trends via Symptom Report, Cognition Scores, and Exercise Levels: A Case Series. Frontiers in Psychiatry. 2019 Sep; 10. PMID: WOS:000487269200001. doi: 10.3389/fpsyt.2019.00652.

74.

Vaidyam

Roux

Torous

. Patient Innovation in Investigating the Effects of Environmental Pollution in Schizophrenia: Case Report of Digital Phenotyping Beyond Apps. Jmir Mental Health. 2020 Aug; 7(8). PMID: WOS:000557334100001. doi: 10.2196/19778.

75.

Rehm

Shield

. Global Burden of Disease and the Impact of Mental and Addictive Disorders. Curr Psychiatry Rep. 2019 Feb 7; 21(2): 10. PMID: 30729322. doi: 10.1007/s11920-019-0997-0.

76.

Cohen

Schwartz

Cowan

Cox

Tucker

, et al. Validating digital phenotyping technologies for clinical use: The critical importance of ‘resolution’. World Psychiatry. 2020; 19(1): 114-5. PMID: 2020-02533-026. doi: 10.1002/wps.20703.

77.

Cohen

Cox

Masucci

Cowan

Coghill

, et al. Digital Phenotyping Using Multimodal Data. Current Behavioral Neuroscience Reports. 2020 Dec; 7(4): 212-20. PMID: WOS:000672000000004. doi: 10.1007/s40473-020-00215-4.

78.

Cohen

Schwartz

Cowan

Kirkpatrick

Raugh

, et al. Digital phenotyping of negative symptoms: the relationship to clinician ratings. Schizophrenia Bulletin. 2021 Jan; 47(1): 44-53. PMID: WOS:000649368900009. doi: 10.1093/schbul/sbaa065.

79.

Torous

Larsen

Depp

Cosco

Barnett

Nock

, et al. Smartphones, Sensors, and Machine Learning to Advance Real-Time Prediction and Interventions for Suicide Prevention: a Review of Current Progress and Next Steps. Current Psychiatry Reports. 2018 Jul; 20(7). PMID: WOS:000436904300001. doi: 10.1007/s11920-018-0914-y.

80.

Ballard

Gilbert

Wusinich

Zarate

. New Methods for Assessing Rapid Changes in Suicide Risk. Frontiers in Psychiatry. 2021 Jan; 12. PMID: WOS:000616095800001. doi: 10.3389/fpsyt.2021.598434.

81.

Allen

Nelson

Brent

Auerbach

. Short-term prediction of suicidal thoughts and behaviors in adolescents: Can recent developments in technology and computational science provide a breakthrough? Journal of Affective Disorders. 2019 May; 250: 163-9. PMID: WOS:000463865400023. doi: 10.1016/j.jad.2019.03.044.

82.

Antosik-Wójcińska

Dominiak

Chojnacka

Kaczmarek-Majer

Opara

Radziszewska

, et al. Smartphone as a monitoring tool for bipolar disorder: a systematic review including data analysis, machine learning algorithms and predictive modelling. Int J Med Inform. 2020 Jun; 138: 104131. PMID: 32305023. doi: 10.1016/j.ijmedinf.2020.104131.

83.

Malhi

Hamilton

Morris

Mannie

Das

Outhred

. The promise of digital mood tracking technologies: are we heading on the right track? Evid Based Ment Health. 2017; Nov; 20(4): 102-7. PMID: 28855245. doi: 10.1136/eb-2017-102757.

84.

Page

McKenzie

Bossuyt

Boutron

Hoffmann

Mulrow

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021; 372: n71. doi: 10.1136/bmj.n71.

Digital phenotyping in psychiatry: A scoping review

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1. Introduction

2. Methods

2.1 Information sources and search strategy

2.2 Inclusion and exclusion criteria

2.3 Data extraction

2.4 Analysis

3.1 Study characteristics

Table 1 Number of studies included and type of study

3.3 Distribution of research among various psychiatric conditions

3.4 Types of data collected

4. Discussion

Data availability

Author contributions

Funding

Supplementary data

Footnotes

Conflict of interest

References

Table 1
Number of studies included and type of study