Toward a Mobile Platform for Real-world Digital Measurement of Depression: User-Centered Design, Data Quality, and Behavioral and Clinical Modeling

Smartphone digital phenotyping in mental health disorders: A review of raw sensors utilized, machine learning processing pipelines, and derived behavioral features

With increased access to digital technology, there has been a surge in the use of and interest in digital phenotyping as a tool to calculate various features from raw smart device data. However, the increased usage of digital phenotyping has created confusion. The vast number of sensors that can be utilized to collect passive data, and diverse methods utilized to convert that sensor data into features has introduced conflicting results and conclusions into the literature. Consequently, there is an identified need for standardizing how digital phenotyping data is measured and collected. This review evaluates the different sensors and methods utilized in digital phenotyping research across 112 papers, with the goal of finding the most common platforms, sensors, and methods for each behavioral measure. This should help guide future digital phenotyping research, and resolve some existing confusion in the field. Information on each study's data sensor variables were tracked and consolidated into a double-coded Codebook. Variables assessed included but were not limited to data sensors, features extracted from data sensors, statistical methods used, phone type, patient access to phones, and characteristics of patient population. This review found that most studies used Android devices (n = 67) or both Android and iPhone (n = 38) with an average duration of 14.3 weeks. The GPS sensor was also found to be the most frequently used sensor. This review underscores the need for standardization in methodological reporting, sensor utilization, and feature extraction across mental health studies.

Investigating Smartphone-Based Sensing Features for Depression Severity Prediction: Observation Study

BACKGROUND

Unobtrusively collected objective sensor data from everyday devices like smartphones provide a novel paradigm to infer mental health symptoms. This process, called smart sensing, allows a fine-grained assessment of various features (eg, time spent at home based on the GPS sensor). Based on its prevalence and impact, depression is a promising target for smart sensing. However, currently, it is unclear which sensor-based features should be used in depression severity prediction and if they hold an incremental benefit over established fine-grained assessments like the ecological momentary assessment (EMA).

OBJECTIVE

The aim of this study was to investigate various features based on the smartphone screen, app usage, and call sensor alongside EMA to infer depression severity. Bivariate, cluster-wise, and cluster-combined analyses were conducted to determine the incremental benefit of smart sensing features compared to each other and EMA in parsimonious regression models for depression severity.

METHODS

In this exploratory observational study, participants were recruited from the general population. Participants needed to be 18 years of age, provide written informed consent, and own an Android-based smartphone. Sensor data and EMA were collected via the INSIGHTS app. Depression severity was assessed using the 8-item Patient Health Questionnaire. Missing data were handled by multiple imputations. Correlation analyses were conducted for bivariate associations; stepwise linear regression analyses were used to find the best prediction models for depression severity. Models were compared by adjusted R2. All analyses were pooled across the imputed datasets according to Rubin's rule.

RESULTS

A total of 107 participants were included in the study. Ages ranged from 18 to 56 (mean 22.81, SD 7.32) years, and 78% of the participants identified as female. Depression severity was subclinical on average (mean 5.82, SD 4.44; Patient Health Questionnaire score ≥10: 18.7%). Small to medium correlations were found for depression severity and EMA (eg, valence: r=-0.55, 95% CI -0.67 to -0.41), and there were small correlations with sensing features (eg, screen duration: r=0.37, 95% CI 0.20 to 0.53). EMA features could explain 35.28% (95% CI 20.73% to 49.64%) of variance and sensing features (adjusted R2=20.45%, 95% CI 7.81% to 35.59%). The best regression model contained EMA and sensing features (R2=45.15%, 95% CI 30.39% to 58.53%).

CONCLUSIONS

Our findings underline the potential of smart sensing and EMA to infer depression severity as isolated paradigms and when combined. Although these could become important parts of clinical decision support systems for depression diagnostics and treatment in the future, confirmatory studies are needed before they can be applied to routine care. Furthermore, privacy, ethical, and acceptance issues need to be addressed.

A template and tutorial for preregistering studies using passive smartphone measures

Passive smartphone measures hold significant potential and are increasingly employed in psychological and biomedical research to capture an individual's behavior. These measures involve the near-continuous and unobtrusive collection of data from smartphones without requiring active input from participants. For example, GPS sensors are used to determine the (social) context of a person, and accelerometers to measure movement. However, utilizing passive smartphone measures presents methodological challenges during data collection and analysis. Researchers must make multiple decisions when working with such measures, which can result in different conclusions. Unfortunately, the transparency of these decision-making processes is often lacking. The implementation of open science practices is only beginning to emerge in digital phenotyping studies and varies widely across studies. Well-intentioned researchers may fail to report on some decisions due to the variety of choices that must be made. To address this issue and enhance reproducibility in digital phenotyping studies, we propose the adoption of preregistration as a way forward. Although there have been some attempts to preregister digital phenotyping studies, a template for registering such studies is currently missing. This could be problematic due to the high level of complexity that requires a well-structured template. Therefore, our objective was to develop a preregistration template that is easy to use and understandable for researchers. Additionally, we explain this template and provide resources to assist researchers in making informed decisions regarding data collection, cleaning, and analysis. Overall, we aim to make researchers' choices explicit, enhance transparency, and elevate the standards for studies utilizing passive smartphone measures.

Innovative Approaches to Addressing Pediatric Mental Health: Digital Technologies in Pediatric Primary Care

Digital technologies can be used at multiple levels to support the mental health care of children including (1) health system/health care provider level; (2) patient-provider interface; (3) patient-facing consumer applications; and (4) new technology, including artificial intelligence. At each of these levels, these novel technologies may lead to care improvements but also may have risks. This review provides an overview of each of innovations across the digital landscape.

The Relation Between Passively Collected GPS Mobility Metrics and Depressive Symptoms: Systematic Review and Meta-Analysis

BACKGROUND

The objective, unobtrusively collected GPS features (eg, homestay and distance) from everyday devices like smartphones may offer a promising augmentation to current assessment tools for depression. However, to date, there is no systematic and meta-analytical evidence on the associations between GPS features and depression.

OBJECTIVE

This study aimed to investigate the between-person and within-person correlations between GPS mobility and activity features and depressive symptoms, and to critically review the quality and potential publication bias in the field.

METHODS

We searched MEDLINE, PsycINFO, Embase, CENTRAL, ACM, IEEE Xplore, PubMed, and Web of Science to identify eligible articles focusing on the correlations between GPS features and depression from December 6, 2022, to March 24, 2023. Inclusion and exclusion criteria were applied in a 2-stage inclusion process conducted by 2 independent reviewers (YT and JK). To be eligible, studies needed to report correlations between wearable-based GPS variables (eg, total distance) and depression symptoms measured with a validated questionnaire. Studies with underage persons and other mental health disorders were excluded. Between- and within-person correlations were analyzed using random effects models. Study quality was determined by comparing studies against the STROBE (Strengthening the Reporting of Observational studies in Epidemiology) guidelines. Publication bias was investigated using Egger test and funnel plots.

RESULTS

A total of k=19 studies involving N=2930 participants were included in the analysis. The mean age was 38.42 (SD 18.96) years with 59.64% (SD 22.99%) of participants being female. Significant between-person correlations between GPS features and depression were identified: distance (r=-0.25, 95% CI -0.29 to -0.21), normalized entropy (r-0.17, 95% CI -0.29 to -0.04), location variance (r-0.17, 95% CI -0.26 to -0.04), entropy (r=-0.13, 95% CI -0.23 to -0.04), number of clusters (r=-0.11, 95% CI -0.18 to -0.03), and homestay (r=0.10, 95% CI 0.00 to 0.19). Studies reporting within-correlations (k=3) were too heterogeneous to conduct meta-analysis. A deficiency in study quality and research standards was identified: all studies followed exploratory observational designs, but no study referenced or fully adhered to the international guidelines for reporting observational studies (STROBE). A total of 79% (k=15) of the studies were underpowered to detect a small correlation (r=.20). Results showed evidence for potential publication bias.

CONCLUSIONS

Our results provide meta-analytical evidence for between-person correlations of GPS mobility and activity features and depression. Hence, depression diagnostics may benefit from adding GPS mobility and activity features as an integral part of future assessment and expert tools. However, confirmatory studies for between-person correlations and further research on within-person correlations are needed. In addition, the methodological quality of the evidence needs to improve.

TRIAL REGISTRATION

OSF Registeries cwder; https://osf.io/cwder.

Identifying a stable and generalizable factor structure of major depressive disorder across three large longitudinal cohorts

The Patient Health Questionnaire 9 (PHQ-9) is the current standard outpatient screening tool for measuring and tracking the nine symptoms of major depressive disorder (MDD). While the PHQ-9 was originally conceptualized as a unidimensional measure, it has become clear that MDD is not a monolithic construct, as evidenced by high comorbidities with other theoretically distinct diagnoses and common symptom overlap between depression and other diagnoses. Therefore, identifying reliable and temporally stable subfactors of depressive symptoms could allow research and care to be tailored to different depression phenotypes. This study improved on previous factor analysis studies of the PHQ-9 by leveraging samples that were clinical (participants with depression only), large (N = 1483 depressed individuals in total), longitudinal (up to 5 years), and from three diverse (matching racial distribution of the United States) datasets. By refraining from assuming the number of factors or item loadings a priori, and thus utilizing a solely data-driven approach, we identified a ranked list of best-fitting models, with the parsimonious one achieving good model fit across studies at most timepoints (average TLI >= 0.90). This model categorizes the PHQ-9 items into four factors: (1) Affective (Anhedonia + Depressed Mood), (2) Somatic (Sleep + Fatigue + Appetite), (3) Internalizing (Worth/Guilt + Suicidality), (4) Sensorimotor (Concentration + Psychomotor), which may be used to further precision psychiatry by testing factor-specific interventions in research and clinical settings.

Digital and precision clinical trials: innovations for testing mental health medications, devices, and psychosocial treatments

Mental health treatment advances - including neuropsychiatric medications and devices, psychotherapies, and cognitive treatments - lag behind other fields of clinical medicine such as cardiovascular care. One reason for this gap is the traditional techniques used in mental health clinical trials, which slow the pace of progress, produce inequities in care, and undermine precision medicine goals. Newer techniques and methodologies, which we term digital and precision trials, offer solutions. These techniques consist of (1) decentralized (i.e., fully-remote) trials which improve the speed and quality of clinical trials and increase equity of access to research, (2) precision measurement which improves success rate and is essential for precision medicine, and (3) digital interventions, which offer increased reach of, and equity of access to, evidence-based treatments. These techniques and their rationales are described in detail, along with challenges and solutions for their utilization. We conclude with a vignette of a depression clinical trial using these techniques.

A Fast and Minimal System to Identify Depression Using Smartphones: Explainable Machine Learning-Based Approach

BACKGROUND

Existing robust, pervasive device-based systems developed in recent years to detect depression require data collected over a long period and may not be effective in cases where early detection is crucial. Additionally, due to the requirement of running systems in the background for prolonged periods, existing systems can be resource inefficient. As a result, these systems can be infeasible in low-resource settings.

OBJECTIVE

Our main objective was to develop a minimalistic system to identify depression using data retrieved in the fastest possible time. Another objective was to explain the machine learning (ML) models that were best for identifying depression.

METHODS

We developed a fast tool that retrieves the past 7 days' app usage data in 1 second (mean 0.31, SD 1.10 seconds). A total of 100 students from Bangladesh participated in our study, and our tool collected their app usage data and responses to the Patient Health Questionnaire-9. To identify depressed and nondepressed students, we developed a diverse set of ML models: linear, tree-based, and neural network-based models. We selected important features using the stable approach, along with 3 main types of feature selection (FS) approaches: filter, wrapper, and embedded methods. We developed and validated the models using the nested cross-validation method. Additionally, we explained the best ML models through the Shapley additive explanations (SHAP) method.

RESULTS

Leveraging only the app usage data retrieved in 1 second, our light gradient boosting machine model used the important features selected by the stable FS approach and correctly identified 82.4% (n=42) of depressed students (precision=75%, F1-score=78.5%). Moreover, after comprehensive exploration, we presented a parsimonious stacking model where around 5 features selected by the all-relevant FS approach Boruta were used in each iteration of validation and showed a maximum precision of 77.4% (balanced accuracy=77.9%). Feature importance analysis suggested app usage behavioral markers containing diurnal usage patterns as being more important than aggregated data-based markers. In addition, a SHAP analysis of our best models presented behavioral markers that were related to depression. For instance, students who were not depressed spent more time on education apps on weekdays, whereas those who were depressed used a higher number of photo and video apps and also had a higher deviation in using photo and video apps over the morning, afternoon, evening, and night time periods of the weekend.

CONCLUSIONS

Due to our system's fast and minimalistic nature, it may make a worthwhile contribution to identifying depression in underdeveloped and developing regions. In addition, our detailed discussion about the implication of our findings can facilitate the development of less resource-intensive systems to better understand students who are depressed and take steps for intervention.

Association Between the Severity of Depressive Symptoms and Human-Smartphone Interactions: Longitudinal Study

BACKGROUND

Various behavioral sensing research studies have found that depressive symptoms are associated with human-smartphone interaction behaviors, including lack of diversity in unique physical locations, entropy of time spent in each location, sleep disruption, session duration, and typing speed. These behavioral measures are often tested against the total score of depressive symptoms, and the recommended practice to disaggregate within- and between-person effects in longitudinal data is often neglected.

OBJECTIVE

We aimed to understand depression as a multidimensional process and explore the association between specific dimensions and behavioral measures computed from passively sensed human-smartphone interactions. We also aimed to highlight the nonergodicity in psychological processes and the importance of disaggregating within- and between-person effects in the analysis.

METHODS

Data used in this study were collected by Mindstrong Health, a telehealth provider that focuses on individuals with serious mental illness. Depressive symptoms were measured by the Diagnostic and Statistical Manual of Mental Disorders Fifth Edition (DSM-5) Self-Rated Level 1 Cross-Cutting Symptom Measure-Adult Survey every 60 days for a year. Participants' interactions with their smartphones were passively recorded, and 5 behavioral measures were developed and were expected to be associated with depressive symptoms according to either theoretical proposition or previous empirical evidence. Multilevel modeling was used to explore the longitudinal relations between the severity of depressive symptoms and these behavioral measures. Furthermore, within- and between-person effects were disaggregated to accommodate the nonergodicity commonly found in psychological processes.

RESULTS

This study included 982 records of DSM Level 1 depressive symptom measurements and corresponding human-smartphone interaction data from 142 participants (age range 29-77 years; mean age 55.1 years, SD 10.8 years; 96 female participants). Loss of interest in pleasurable activities was associated with app count (γ₁₀=-0.14; P=.01; within-person effect). Depressed mood was associated with typing time interval (γ₀₅=0.88; P=.047; within-person effect) and session duration (γ₀₅=-0.37; P=.03; between-person effect).

CONCLUSIONS

This study contributes new evidence for associations between human-smartphone interaction behaviors and the severity of depressive symptoms from a dimensional perspective, and it highlights the importance of considering the nonergodicity of psychological processes and analyzing the within- and between-person effects separately.

Relapse prediction in schizophrenia with smartphone digital phenotyping during COVID-19: a prospective, three-site, two-country, longitudinal study

Smartphone technology provides us with a more convenient and less intrusive method of detecting changes in behavior and symptoms that typically precede schizophrenia relapse. To take advantage of the aforementioned, this study examines the feasibility of predicting schizophrenia relapse by identifying statistically significant anomalies in patient data gathered through mindLAMP, an open-source smartphone app. Participants, recruited in Boston, MA in the United States, and Bangalore and Bhopal in India, were invited to use mindLAMP for up to a year. The passive data (geolocation, accelerometer, and screen state), active data (surveys), and data quality metrics collected by the app were then retroactively fed into a relapse prediction model that utilizes anomaly detection. Overall, anomalies were 2.12 times more frequent in the month preceding a relapse and 2.78 times more frequent in the month preceding and following a relapse compared to intervals without relapses. The anomaly detection model incorporating passive data proved a better predictor of relapse than a naive model utilizing only survey data. These results demonstrate that relapse prediction models utilizing patient data gathered by a smartphone app can warn the clinician and patient of a potential schizophrenia relapse.

Applications of Speech Analysis in Psychiatry

ABSTRACT

The need for objective measurement in psychiatry has stimulated interest in alternative indicators of the presence and severity of illness. Speech may offer a source of information that bridges the subjective and objective in the assessment of mental disorders. We systematically reviewed the literature for articles exploring speech analysis for psychiatric applications. The utility of speech analysis depends on how accurately speech features represent clinical symptoms within and across disorders. We identified four domains of the application of speech analysis in the literature: diagnostic classification, assessment of illness severity, prediction of onset of illness, and prognosis and treatment outcomes. We discuss the findings in each of these domains, with a focus on how types of speech features characterize different aspects of psychopathology. Models that bring together multiple speech features can distinguish speakers with psychiatric disorders from healthy controls with high accuracy. Differentiating between types of mental disorders and symptom dimensions are more complex problems that expose the transdiagnostic nature of speech features. Convergent progress in speech research and computer sciences opens avenues for implementing speech analysis to enhance objectivity of assessment in clinical practice. Application of speech analysis will need to address issues of ethics and equity, including the potential to perpetuate discriminatory bias through models that learn from clinical assessment data. Methods that mitigate bias are available and should play a key role in the implementation of speech analysis.

Smartphone accelerometer data as a proxy for clinical data in modeling of bipolar disorder symptom trajectory

Being able to track and predict fluctuations in symptoms of mental health disorders such as bipolar disorder outside the clinic walls is critical for expanding access to care for the global population. To that end, we analyze a dataset of 291 individuals from a smartphone app targeted at bipolar disorder, which contains rich details about their smartphone interactions (including typing dynamics and accelerometer motion) collected everyday over several months, along with more traditional clinical features. The aim is to evaluate whether smartphone accelerometer data could serve as a proxy for traditional clinical data, either by itself or in combination with typing dynamics. Results show that accelerometer data improves the predictive performance of machine learning models by nearly 5% over those previously reported in the literature based only on clinical data and typing dynamics. This suggests it is possible to elicit essentially the same "information" about bipolar symptomology using different data sources, in a variety of settings.

Mindfulness Training for Depressed Older Adults Using Smartphone Technology: Protocol for a Fully Remote Precision Clinical Trial

BACKGROUND

Precision medicine, optimized interventions, and access to care are catchphrases for the future of behavioral treatments. Progress has been slow due to the dearth of clinical trials that optimize interventions' benefits, individually tailor interventions to meet individual needs and preferences, and lead to rapid implementation after effectiveness is demonstrated. Two innovations have emerged to meet these challenges: fully remote trials and precision clinical trials.

OBJECTIVE

This paper provides a detailed description of Mindful MyWay, a study designed to test online mindfulness training in older adults with depression. Consistent with the concept of fully remote trials using a smartphone app, the study requires no in-person contact and can be conducted with participants anywhere in the United States. Based upon the precision medicine framework, the study assesses participants using high-frequency assessments of symptoms, cognitive performance, and patient preferences to both understand the individualized nature of treatment response and help individually tailor the intervention.

METHODS

Mindful MyWay is an open-label early-phase clinical trial for individuals 65 years and older with current depression. A smartphone app was developed to help coordinate the study, deliver the intervention, and evaluate the acceptability of the intervention, as well as predictors and outcomes of it. The curriculum for the fully remote intervention parallels the mindfulness-based stress reduction curriculum, a protocolized group-based mindfulness training that is typically provided in person. After consent and screening, participants download The Healthy Mind Lab mobile health smartphone app from the Apple App Store, allowing them to complete brief smartphone-based assessments of depressive symptoms and cognitive performance 4 times each day for 4 weeks prior to and after completing the intervention. The intervention consists of an introduction video and 10 weekly mindfulness training sessions, with the expectation to practice mindfulness at home daily. The app collects participant preference data throughout the 10-week intervention period; these high-frequency assessments identify participants' individually dynamic preferences toward the goal of optimizing the intervention in future iterations.

RESULTS

Participant recruitment and data collection began in March 2019. Final end point assessments will be collected in May 2022. The paper describes lessons learned regarding the critical role of early-phase testing prior to moving to a randomized trial.

CONCLUSIONS

The Mindful MyWay study is an exemplar of innovative clinical trial designs that use smartphone technology in behavioral and neuropsychiatric conditions. These include fully remote studies that can recruit throughout the United States, including hard-to-access areas, and collect high-frequency data, which is ideal for idiographic assessment and individualized intervention optimization. Our findings will be used to modify our methods and inform future randomized controlled trials within a precision medicine framework.

TRIAL REGISTRATION

ClinicalTrials.gov NCT03922217; https://clinicaltrials.gov/ct2/show/NCT03922217.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

DERR1-10.2196/39233.

Participant retention in a fully remote trial of digital psychotherapy: Comparison of incentive types

Numerous studies have found that long term retention is very low in remote clinical studies (>4 weeks) and to date there is limited information on the best methods to ensure retention. The ability to retain participants in the completion of key assessments periods is critical to all clinical research, and to date little is known as to what methods are best to encourage participant retention. To study incentive-based retention methods we randomized 215 US adults (18+ years) who agreed to participate in a sequential, multiple assignment randomized trial to either high monetary incentive (HMI, $125 USD) and combined low monetary incentive ($75 USD) plus alternative incentive (LMAI). Participants were asked to complete daily and weekly surveys for a total of 12 weeks, which included a tailoring assessment around week 5 to determine who should be stepped up and rerandomized to one of two augmentation conditions. Key assessment points were weeks 5 and 12. There was no difference in participant retention at week 5 (tailoring event), with approximately 75% of the sample completing the week-5 survey. By week 10, the HMI condition retained approximately 70% of the sample, compared to 60% of the LMAI group. By week 12, all differences were attenuated. Differences in completed measures were not significant between groups. At the end of the study, participants were asked the impressions of the incentive condition they were assigned and asked for suggestions for improving engagement. There were no significant differences between conditions on ratings of the fairness of compensation, study satisfaction, or study burden, but study burden, intrinsic motivation and incentive fairness did influence participation. Men were also more likely to drop out of the study than women. Qualitative analysis from both groups found the following engagement suggestions: desire for feedback on survey responses and an interest in automated sharing of individual survey responses with study therapists to assist in treatment. Participants in the LMAI arm indicated that the alternative incentives were engaging and motivating. In sum, while we were able to increase engagement above what is typical for such study, more research is needed to truly improve long term retention in remote trials.

Recruitment & Retention in Remote Research: Learnings from a Large Decentralized Real-World Study (Preprint)

BACKGROUND

Smartphones are increasingly used in health research. They provide a continuous connection between participants and researchers to monitor long-term health trajectories of large populations at a fraction of the cost of traditional research studies. However, despite the potential of using smartphones in remote research, there is an urgent need to develop effective strategies to reach, recruit, and retain the target populations in a representative and equitable manner.

OBJECTIVE

We aimed to investigate the impact of combining different recruitment and incentive distribution approaches used in remote research on cohort characteristics and long-term retention. The real-world factors significantly impacting active and passive data collection were also evaluated.

METHODS

We conducted a secondary data analysis of participant recruitment and retention using data from a large remote observation study aimed at understanding real-world factors linked to cold, influenza, and the impact of traumatic brain injury on daily functioning. We conducted recruitment in 2 phases between March 15, 2020, and January 4, 2022. Over 10,000 smartphone owners in the United States were recruited to provide 12 weeks of daily surveys and smartphone-based passive-sensing data. Using multivariate statistics, we investigated the potential impact of different recruitment and incentive distribution approaches on cohort characteristics. Survival analysis was used to assess the effects of sociodemographic characteristics on participant retention across the 2 recruitment phases. Associations between passive data-sharing patterns and demographic characteristics of the cohort were evaluated using logistic regression.

RESULTS

We analyzed over 330,000 days of engagement data collected from 10,000 participants. Our key findings are as follows: first, the overall characteristics of participants recruited using digital advertisements on social media and news media differed significantly from those of participants recruited using crowdsourcing platforms (Prolific and Amazon Mechanical Turk; P<.001). Second, participant retention in the study varied significantly across study phases, recruitment sources, and socioeconomic and demographic factors (P<.001). Third, notable differences in passive data collection were associated with device type (Android vs iOS) and participants' sociodemographic characteristics. Black or African American participants were significantly less likely to share passive sensor data streams than non-Hispanic White participants (odds ratio 0.44-0.49, 95% CI 0.35-0.61; P<.001). Fourth, participants were more likely to adhere to baseline surveys if the surveys were administered immediately after enrollment. Fifth, technical glitches could significantly impact real-world data collection in remote settings, which can severely impact generation of reliable evidence.

CONCLUSIONS

Our findings highlight several factors, such as recruitment platforms, incentive distribution frequency, the timing of baseline surveys, device heterogeneity, and technical glitches in data collection infrastructure, that could impact remote long-term data collection. Combined together, these empirical findings could help inform best practices for monitoring anomalies during real-world data collection and for recruiting and retaining target populations in a representative and equitable manner.

Digital phenotyping correlations in larger mental health samples: analysis and replication

BACKGROUND

Smartphones can facilitate patients completing surveys and collecting sensor data to gain insight into their mental health conditions. However, the utility of sensor data is still being explored. Prior studies have reported a wide range of correlations between passive data and survey scores.

AIMS

To explore correlations in a large data-set collected with the mindLAMP app. Additionally, we explored whether passive data features could be used in models to predict survey results.

METHOD

Participants were asked to complete daily and weekly mental health surveys. After screening for data quality, our sample included 147 college student participants and 270 weeks of data. We examined correlations between six weekly surveys and 13 metrics derived from passive data features. Finally, we trained logistic regression models to predict survey scores from passive data with and without daily surveys.

RESULTS

Similar to other large studies, our correlations were lower than prior reports from smaller studies. We found that the most useful features came from GPS, call, and sleep duration data. Logistic regression models performed poorly with only passive data, but when daily survey scores were included, performance greatly increased.

CONCLUSIONS

Although passive data alone may not provide enough information to predict survey scores, augmenting this data with short daily surveys can improve performance. Therefore, it may be that passive data can be used to refine survey score predictions and clinical utility may be derived from the combination of active and passive data.

Similarity matrix-based anomaly detection for clinical intervention

The use of digital phenotyping methods in clinical care has allowed for improved investigation of spatiotemporal behaviors of patients. Moreover, detecting abnormalities in mobile sensor data patterns can be instrumental in identifying potential changes in symptomology. We propose a method that temporally aligns sensor data in order to achieve interpretable measures of similarity between time points. These computed measures can then be used for anomaly detection, baseline routine computation, and trajectory clustering. In addition, we apply this method on a study of 695 college participants, as well as on a patient with worsening anxiety and depression. With varying temporal constraints, we find mild correlations between changes in routine and clinical scores. Furthermore, in our experiment on an individual with elevated depression and anxiety, we are able to cluster GPS trajectories, allowing for improved understanding and visualization of routines with respect to symptomology. In the future, we aim to apply this method on individuals that undergo data collection for longer periods of time, thus allowing for a better understanding of long-term routines and signals for clinical intervention.

Digital phenotype of mood disorders: A conceptual and critical review

BACKGROUND

Mood disorders are commonly diagnosed and staged using clinical features that rely merely on subjective data. The concept of digital phenotyping is based on the idea that collecting real-time markers of human behavior allows us to determine the digital signature of a pathology. This strategy assumes that behaviors are quantifiable from data extracted and analyzed through digital sensors, wearable devices, or smartphones. That concept could bring a shift in the diagnosis of mood disorders, introducing for the first time additional examinations on psychiatric routine care.

OBJECTIVE

The main objective of this review was to propose a conceptual and critical review of the literature regarding the theoretical and technical principles of the digital phenotypes applied to mood disorders.

METHODS

We conducted a review of the literature by updating a previous article and querying the PubMed database between February 2017 and November 2021 on titles with relevant keywords regarding digital phenotyping, mood disorders and artificial intelligence.

RESULTS

Out of 884 articles included for evaluation, 45 articles were taken into account and classified by data source (multimodal, actigraphy, ECG, smartphone use, voice analysis, or body temperature). For depressive episodes, the main finding is a decrease in terms of functional and biological parameters [decrease in activities and walking, decrease in the number of calls and SMS messages, decrease in temperature and heart rate variability (HRV)], while the manic phase produces the reverse phenomenon (increase in activities, number of calls and HRV).

CONCLUSION

The various studies presented support the potential interest in digital phenotyping to computerize the clinical characteristics of mood disorders.