Smartphone-Based Passive Assessment of Mobility in Depression: Challenges and Opportunities

How Could Sensor-Based Measurement of Physical Activity Be Used in Cardiovascular Healthcare?

Physical activity and cardiovascular disease (CVD) are intimately linked. Low levels of physical activity increase the risk of CVDs, including myocardial infarction and stroke. Conversely, when CVD develops, it often reduces the ability to be physically active. Despite these largely understood relationships, the objective measurement of physical activity is rarely performed in routine healthcare. The ability to use sensor-based approaches to accurately measure aspects of physical activity has the potential to improve many aspects of cardiovascular healthcare across the spectrum of healthcare, from prediction, prevention, diagnosis, and treatment to disease monitoring. This review discusses the potential of sensor-based measurement of physical activity to augment current cardiovascular healthcare. We highlight many factors that should be considered to maximise the benefit and reduce the risks of such an approach. Because the widespread use of such devices in society is already a reality, it is important that scientists, clinicians, and healthcare providers are aware of these considerations.

Exploratory Machine Learning Modeling of Adaptive and Maladaptive Personality Traits from Passively Sensed Behavior

Continuous passive sensing of daily behavior from mobile devices has the potential to identify behavioral patterns associated with different aspects of human characteristics. This paper presents novel analytic approaches to extract and understand these behavioral patterns and their impact on predicting adaptive and maladaptive personality traits. Our machine learning analysis extends previous research by showing that both adaptive and maladaptive traits are associated with passively sensed behavior providing initial evidence for the utility of this type of data to study personality and its pathology. The analysis also suggests directions for future confirmatory studies into the underlying behavior patterns that link adaptive and maladaptive variants consistent with contemporary models of personality pathology.

MindKind: A mixed-methods protocol for the feasibility of global digital mental health studies in young people

Background: While an estimated 14-20% of young adults experience mental health conditions worldwide, the best strategies for prevention and management are not fully understood. The ubiquity of smartphone use among young people makes them excellent candidates for collecting data about lived experiences and their relationships to mental health. However, not much is known about the factors affecting young peoples’ willingness to share information about their mental health.

Objective: We aim to understand the data governance and engagement strategies influencing young peoples’ (aged 16-24) participation in app-based studies of mental health. We hypothesize that willingness to participate in research is influenced by involvement  in how their data is collected, shared, and used.

Methods: Here, we describe the MindKind Study, which employs mixed methods to understand the feasibility of global, smartphone-based studies of youth mental health. A pilot 12-week app-based substudy will query participants’ willingness to engage with remote mental health studies. Participants will be randomized into one of four different data governance models designed to understand their preferences, as well as the acceptability of models that allow them more or less control over how their data are accessed and used. Enrolees will receive one of two different engagement strategies. A companion qualitative study will employ a deliberative democracy approach to examine the preferences, concerns and expectations of young people, with respect to remote mental health research. We also detail our engagement with young people as co-researchers in this study. This pilot study is being conducted in India, South Africa and the United Kingdom.

Conclusions: This study is expected to generate new insights into the feasibility of, and best practices for, remote smartphone-based studies of mental health in youth and represents an important step toward understanding which approaches could help people better manage their mental health.

Using digital health tools for the Remote Assessment of Treatment Prognosis in Depression (RAPID): a study protocol for a feasibility study

INTRODUCTION

Digital health tools such as smartphones and wearable devices could improve psychological treatment outcomes in depression through more accurate and comprehensive measures of patient behaviour. However, in this emerging field, most studies are small and based on student populations outside of a clinical setting. The current study aims to determine the feasibility and acceptability of using smartphones and wearable devices to collect behavioural and clinical data in people undergoing therapy for depressive disorders and establish the extent to which they can be potentially useful biomarkers of depression and recovery after treatment.

METHODS AND ANALYSIS

This is an observational, prospective cohort study of 65 people attending psychological therapy for depression in multiple London-based sites. It will collect continuous passive data from smartphone sensors and a Fitbit fitness tracker, and deliver questionnaires, speech tasks and cognitive assessments through smartphone-based apps. Objective data on sleep, physical activity, location, Bluetooth contact, smartphone use and heart rate will be gathered for 7 months, and compared with clinical and contextual data. A mixed methods design, including a qualitative interview of patient experiences, will be used to evaluate key feasibility indicators, digital phenotypes of depression and therapy prognosis. Patient and public involvement was sought for participant-facing documents and the study design of the current research proposal.

ETHICS AND DISSEMINATION

Ethical approval has been obtained from the London Westminster Research Ethics Committee, and the Health Research Authority, Integrated Research Application System (project ID: 270918). Privacy and confidentiality will be guaranteed and the procedures for handling, processing, storage and destruction of the data will comply with the General Data Protection Regulation. Findings from this study will form part of a doctoral thesis, will be presented at national and international meetings or academic conferences and will generate manuscripts to be submitted to peer-reviewed journals.

TRIAL REGISTRATION NUMBER

https://doi.org/10.17605/OSF.IO/PMYTA.

Predicting Changes in Depression Severity Using the PSYCHE-D (Prediction of Severity Change-Depression) Model Involving Person-Generated Health Data: Longitudinal Case-Control Observational Study

Background

In 2017, an estimated 17.3 million adults in the United States experienced at least one major depressive episode, with 35% of them not receiving any treatment. Underdiagnosis of depression has been attributed to many reasons, including stigma surrounding mental health, limited access to medical care, and barriers due to cost.

Objective

This study aimed to determine if low-burden personal health solutions, leveraging person-generated health data (PGHD), could represent a possible way to increase engagement and improve outcomes.

Methods

Here, we present the development of PSYCHE-D (Prediction of Severity Change-Depression), a predictive model developed using PGHD from more than 4000 individuals, which forecasts the long-term increase in depression severity. PSYCHE-D uses a 2-phase approach. The first phase supplements self-reports with intermediate generated labels, and the second phase predicts changing status over a 3-month period, up to 2 months in advance. The 2 phases are implemented as a single pipeline in order to eliminate data leakage and ensure results are generalizable.

Results

PSYCHE-D is composed of 2 Light Gradient Boosting Machine (LightGBM) algorithm–based classifiers that use a range of PGHD input features, including objective activity and sleep, self-reported changes in lifestyle and medication, and generated intermediate observations of depression status. The approach generalizes to previously unseen participants to detect an increase in depression severity over a 3-month interval, with a sensitivity of 55.4% and a specificity of 65.3%, nearly tripling sensitivity while maintaining specificity when compared with a random model.

Conclusions

These results demonstrate that low-burden PGHD can be the basis of accurate and timely warnings that an individual’s mental health may be deteriorating. We hope this work will serve as a basis for improved engagement and treatment of individuals experiencing depression.

Continuous TBI monitoring from Spontaneous Speech using Parametrized Sinc Filters and a Cascading GRU

Traumatic Brain Injury (TBI) is caused by a head injury that affects the brain, impairing cognitive and communication function and resulting in speech and language disorders. Over 80,000 individuals in the US suffer from long-term TBI disabilities and continuous monitoring after TBI is essential to facilitate rehabilitation and prevent regression. Prior work has demonstrated the feasibility of TBI monitoring from speech by leveraging advancements in Artificial Intelligence (AI) and speech processing technology. However, much of prior work explored TBI detection using audio captured using a mobile device while subjects performed scripted speech tasks such as diadochokinesis tests or read a passage. Such scripted approaches require active user involvement that significantly burdens participants. Moreover, they are episodic and do not provide a longitudinal picture of the user's TBI condition, which is useful in monitoring recovery trajectory. This study proposes a continuous TBI monitoring from changes in acoustic features of spontaneous speech collected passively using the smartphone. Low-level acoustic features are extracted using parametrized Sinc filters (pSinc) that are then classified TBI (yes/no) using a cascading Gated Recurrent Unit (cGRU). The cGRU model utilizes a cell gate unit in the GRU to store and incorporate each individual's prediction history as prior knowledge into the model. In rigorous evaluation, our proposed method outperformed prior TBI detection methods on a dataset containing conversational speech recorded during patient-therapist discourses following TBI, achieving 83.87% balanced TBI classification accuracy. Furthermore, unique words that are important in TBI prediction were identified using SHapley Additive exPlanations (SHAP). A correlation was also found between features acquired by the proposed method and coordination deficits following TBI.

MindKind: A mixed-methods protocol for the feasibility of global digital mental health studies in young people

Background: While an estimated 14-20% of young adults experience mental health conditions worldwide, the best strategies for prevention and management are not fully understood. The ubiquity of smartphone use among young people makes them excellent candidates for collecting data about lived experiences and their relationships to mental health. However, not much is known about the factors affecting young peoples’ willingness to share information about their mental health. Objective: We aim to understand the data governance and engagement strategies influencing young peoples’ (aged 16-24) participation in app-based studies of mental health. We hypothesize that the willingness to participate in research is impacted by their ability to be involved in how their data is collected, shared, and used. Methods: Here, we describe the MindKind Study, which employs mixed methods to understand the feasibility of global, smartphone-based studies of youth mental health. A pilot 12-week app-based substudy will query participants’ willingness to engage with remote mental health studies. Participants will be randomized into one of four different data governance models designed to understand their preferences, as well as the acceptability of models that allow them more or less control over how their data are accessed and used. Enrolees will receive one of two different engagement strategies. A companion qualitative study will employ a deliberative democracy approach to examine the preferences, concerns and expectations of young people, with respect to remote mental health research. We also detail our engagement with young people as co-researchers in this study. This pilot study is being conducted in India, South Africa and the United Kingdom. Conclusions: This study is expected to generate new insights into the feasibility of, and best practices for, remote smartphone-based studies of mental health in youth and represents an important step toward understanding which approaches could help people better manage their mental health.

Decision Models and Technology Can Help Psychiatry Develop Biomarkers

Why is psychiatry unable to define clinically useful biomarkers? We explore this question from the vantage of data and decision science and consider biomarkers as a form of phenotypic data that resolves a well-defined clinical decision. We introduce a framework that systematizes different forms of phenotypic data and further introduce the concept of decision model to describe the strategies a clinician uses to seek out, combine, and act on clinical data. Though many medical specialties rely on quantitative clinical data and operationalized decision models, we observe that, in psychiatry, clinical data are gathered and used in idiosyncratic decision models that exist solely in the clinician's mind and therefore are outside empirical evaluation. This, we argue, is a fundamental reason why psychiatry is unable to define clinically useful biomarkers: because psychiatry does not currently quantify clinical data, decision models cannot be operationalized and, in the absence of an operationalized decision model, it is impossible to define how a biomarker might be of use. Here, psychiatry might benefit from digital technologies that have recently emerged specifically to quantify clinically relevant facets of human behavior. We propose that digital tools might help psychiatry in two ways: first, by quantifying data already present in the standard clinical interaction and by allowing decision models to be operationalized and evaluated; second, by testing whether new forms of data might have value within an operationalized decision model. We reference successes from other medical specialties to illustrate how quantitative data and operationalized decision models improve patient care.

Artificial Intelligence: An Interprofessional Perspective on Implications for Geriatric Mental Health Research and Care

Artificial intelligence (AI) in healthcare aims to learn patterns in large multimodal datasets within and across individuals. These patterns may either improve understanding of current clinical status or predict a future outcome. AI holds the potential to revolutionize geriatric mental health care and research by supporting diagnosis, treatment, and clinical decision-making. However, much of this momentum is driven by data and computer scientists and engineers and runs the risk of being disconnected from pragmatic issues in clinical practice. This interprofessional perspective bridges the experiences of clinical scientists and data science. We provide a brief overview of AI with the main focus on possible applications and challenges of using AI-based approaches for research and clinical care in geriatric mental health. We suggest future AI applications in geriatric mental health consider pragmatic considerations of clinical practice, methodological differences between data and clinical science, and address issues of ethics, privacy, and trust.

Assessing the relationship between routine and schizophrenia symptoms with passively sensed measures of behavioral stability

Increased stability in one's daily routine is associated with well-being in the general population and often a goal of behavioral interventions for people with serious mental illnesses like schizophrenia. Assessing behavioral stability has been limited in clinical research by the use of retrospective scales, which are susceptible to reporting biases and memory inaccuracies. Mobile passive sensors, which are less susceptible to these sources of error, have emerged as tools to assess behavioral patterns in a range of populations. The present study developed and examined a metric of behavioral stability from data generated by a passive sensing system carried by 61 individuals with schizophrenia for one year. This metric-the Stability Index-appeared orthogonal from existing measures drawn from passive sensors and matched the predictive performance of state-of-the-art features. Specifically, greater stability in social activity (e.g., calls and messages) were associated with lower symptoms, and greater stability in physical activity (e.g., being still) appeared associated with elevated symptoms. This study provides additional support for the predictive value of individualized over population-level data in psychiatric populations. The Stability Index offers also a promising tool for generating insights about the impact of behavioral stability in schizophrenia-spectrum disorders.

Towards clinically actionable digital phenotyping targets in schizophrenia

Digital phenotyping has potential to quantify the lived experience of mental illness and generate real-time, actionable results related to recovery, such as the case of social rhythms in individuals with bipolar disorder. However, passive data features for social rhythm clinical targets in individuals with schizophrenia have yet to be studied. In this paper, we explore the relationship between active and passive data by focusing on temporal stability and variance at an individual level as well as large-scale associations on a population level to gain clinically actionable information regarding social rhythms. From individual data clustering, we found a 19% cluster overlap between specific active and passive data features for participants with schizophrenia. In the same clinical population, two passive data features in particular associated with social rhythms, "Circadian Routine" and "Weekend Day Routine," and were negatively associated with symptoms of anxiety, depression, psychosis, and poor sleep (Spearman ρ ranged from -0.23 to -0.30, p < 0.001). Conversely, in healthy controls, more stable social rhythms were positively correlated with symptomatology (Spearman ρ ranged from 0.20 to 0.44, p < 0.05). Our results suggest that digital phenotyping in schizophrenia may offer clinically relevant information for understanding how daily routines affect symptomatology. Specifically, negative correlations between smartphone reported anxiety, depression, psychosis, and poor sleep in individuals with schizophrenia, but not in healthy controls, offer an actionable clinical target and area for further investigation.

Mobile Phone and Wearable Sensor-Based mHealth Approaches for Psychiatric Disorders and Symptoms: Systematic Review

BACKGROUND

Mobile Therapeutic Attention for Patients with Treatment-Resistant Schizophrenia (m-RESIST) is an EU Horizon 2020-funded project aimed at designing and validating an innovative therapeutic program for treatment-resistant schizophrenia. The program exploits information from mobile phones and wearable sensors for behavioral tracking to support intervention administration.

OBJECTIVE

To systematically review original studies on sensor-based mHealth apps aimed at uncovering associations between sensor data and symptoms of psychiatric disorders in order to support the m-RESIST approach to assess effectiveness of behavioral monitoring in therapy.

METHODS

A systematic review of the English-language literature, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, was performed through Scopus, PubMed, Web of Science, and the Cochrane Central Register of Controlled Trials databases. Studies published between September 1, 2009, and September 30, 2018, were selected. Boolean search operators with an iterative combination of search terms were applied.

RESULTS

Studies reporting quantitative information on data collected from mobile use and/or wearable sensors, and where that information was associated with clinical outcomes, were included. A total of 35 studies were identified; most of them investigated bipolar disorders, depression, depression symptoms, stress, and symptoms of stress, while only a few studies addressed persons with schizophrenia. The data from sensors were associated with symptoms of schizophrenia, bipolar disorders, and depression.

CONCLUSIONS

Although the data from sensors demonstrated an association with the symptoms of schizophrenia, bipolar disorders, and depression, their usability in clinical settings to support therapeutic intervention is not yet fully assessed and needs to be scrutinized more thoroughly.

The accuracy of passive phone sensors in predicting daily mood

BACKGROUND

Smartphones provide a low-cost and efficient means to collect population level data. Several small studies have shown promise in predicting mood variability from smartphone-based sensor and usage data, but have not been generalized to nationally recruited samples. This study used passive smartphone data, demographic characteristics, and baseline depressive symptoms to predict prospective daily mood.

METHOD

Daily phone usage data were collected passively from 271 Android phone users participating in a fully remote randomized controlled trial of depression treatment (BRIGHTEN). Participants completed daily Patient Health Questionnaire-2. A machine learning approach was used to predict daily mood for the entire sample and individual participants.

RESULTS

Sample-wide estimates showed a marginally significant association between physical mobility and self-reported daily mood (B = -0.04, P < 0.05), but the predictive models performed poorly for the sample as a whole (median R² ∼ 0). Focusing on individuals, 13.9% of participants showed significant association (FDR < 0.10) between a passive feature and daily mood. Personalized models combining features provided better prediction performance (median area under the curve [AUC] > 0.50) for 80.6% of participants and very strong prediction in a subset (median AUC > 0.80) for 11.8% of participants.

CONCLUSIONS

Passive smartphone data with current features may not be suited for predicting daily mood at a population level because of the high degree of intra- and interindividual variation in phone usage patterns and daily mood ratings. Personalized models show encouraging early signs for predicting an individual's mood state changes, with GPS-derived mobility being the top most important feature in the present sample.