Depression predictions from GPS-based mobility do not generalize well to large demographically heterogeneous samples

Mobile sensing-based depression severity assessment in participants with heterogeneous mental health conditions

Mobile sensing-based depression severity assessment could complement the subjective questionnaires-based assessment currently used in practice. However, previous studies on mobile sensing for depression severity assessment were conducted on homogeneous mental health condition participants; evaluation of possible generalization across heterogeneous groups has been limited. Similarly, previous studies have not investigated the potential of free-living audio data for depression severity assessment. Audio recordings from free-living could provide rich sociability features to characterize depressive states. We conducted a study with 11 healthy individuals, 13 individuals with major depressive disorder, and eight individuals with schizoaffective disorders. Communication logs and location data from the participants' smartphones and continuous audio recordings of free-living from a wearable audioband were obtained over a week for each participant. The depression severity prediction model trained using communication log and location data features had a root mean squared error (rmse) of 6.80. Audio-based sociability features further reduced the rmse to 6.07 (normalized rmse of 0.22). Audio-based sociability features also improved the F1 score in the five-class depression category classification model from 0.34 to 0.46. Thus, free-living audio-based sociability features complement the commonly used mobile sensing features to improve depression severity assessment. The prediction results obtained with mobile sensing-based features are better than the rmse of 9.83 (normalized rmse of 0.36) and the F1 score of 0.25 obtained with a baseline model. Additionally, the predicted depression severity had a significant correlation with reported depression severity (correlation coefficient of 0.76, p < 0.001). Thus, our work shows that mobile sensing could model depression severity across participants with heterogeneous mental health conditions, potentially offering a screening tool for depressive symptoms monitoring in the broader population.

Digital biomarkers in depression: A systematic review and call for standardization and harmonization of feature engineering

BACKGROUND

General physicians misclassify depression in more than half of the cases. Researchers have explored the feasibility of leveraging passively collected data points, also called digital biomarkers, to provide more granular understanding of depression phenotypes as well as a more objective assessment of disease.

METHOD

This paper provides a systematic review following the PRISMA guidelines (Page et al., 2021) to understand which digital biomarkers might be relevant for passive screening of depression. Pubmed and PsycInfo were systematically searched for studies published from 2019 to early 2024, resulting in 161 records assessed for eligibility. Excluded were intervention studies, studies focusing on a different disease or those with a lack of passive data collection. 74 studies remained for a quality assessment, after which 27 studies were included.

RESULTS

The review shows that depressed participants' real-life behavior such as reduced communication with others can be tracked by passive data. Machine learning models for the classification of depression have shown accuracies up to 0.98, surpassing the quality of many standardized assessment methods.

LIMITATIONS

Inconsistency of outcome reporting of current studies does not allow for drawing statistical conclusions regarding effectiveness of individual included features. The Covid-19 pandemic might have impacted the ongoing studies between 2020 and 2022.

CONCLUSION

While digital biomarkers allow real-life tracking of participant's behavior and symptoms, further work is required to align the feature engineering of digital biomarkers. With shown high accuracies of assessments, connecting digital biomarkers with clinical practice can be a promising method of detecting symptoms of depression automatically.

Framework for Ranking Machine Learning Predictions of Limited, Multimodal, and Longitudinal Behavioral Passive Sensing Data: Combining User-Agnostic and Personalized Modeling

BACKGROUND

Passive mobile sensing provides opportunities for measuring and monitoring health status in the wild and outside of clinics. However, longitudinal, multimodal mobile sensor data can be small, noisy, and incomplete. This makes processing, modeling, and prediction of these data challenging. The small size of the data set restricts it from being modeled using complex deep learning networks. The current state of the art (SOTA) tackles small sensor data sets following a singular modeling paradigm based on traditional machine learning (ML) algorithms. These opt for either a user-agnostic modeling approach, making the model susceptible to a larger degree of noise, or a personalized approach, where training on individual data alludes to a more limited data set, giving rise to overfitting, therefore, ultimately, having to seek a trade-off by choosing 1 of the 2 modeling approaches to reach predictions.

OBJECTIVE

The objective of this study was to filter, rank, and output the best predictions for small, multimodal, longitudinal sensor data using a framework that is designed to tackle data sets that are limited in size (particularly targeting health studies that use passive multimodal sensors) and that combines both user agnostic and personalized approaches, along with a combination of ranking strategies to filter predictions.

METHODS

In this paper, we introduced a novel ranking framework for longitudinal multimodal sensors (FLMS) to address challenges encountered in health studies involving passive multimodal sensors. Using the FLMS, we (1) built a tensor-based aggregation and ranking strategy for final interpretation, (2) processed various combinations of sensor fusions, and (3) balanced user-agnostic and personalized modeling approaches with appropriate cross-validation strategies. The performance of the FLMS was validated with the help of a real data set of adolescents diagnosed with major depressive disorder for the prediction of change in depression in the adolescent participants.

RESULTS

Predictions output by the proposed FLMS achieved a 7% increase in accuracy and a 13% increase in recall for the real data set. Experiments with existing SOTA ML algorithms showed an 11% increase in accuracy for the depression data set and how overfitting and sparsity were handled.

CONCLUSIONS

The FLMS aims to fill the gap that currently exists when modeling passive sensor data with a small number of data points. It achieves this through leveraging both user-agnostic and personalized modeling techniques in tandem with an effective ranking strategy to filter predictions.

Measuring algorithmic bias to analyze the reliability of AI tools that predict depression risk using smartphone sensed-behavioral data

AI tools intend to transform mental healthcare by providing remote estimates of depression risk using behavioral data collected by sensors embedded in smartphones. While these tools accurately predict elevated depression symptoms in small, homogenous populations, recent studies show that these tools are less accurate in larger, more diverse populations. In this work, we show that accuracy is reduced because sensed-behaviors are unreliable predictors of depression across individuals: sensed-behaviors that predict depression risk are inconsistent across demographic and socioeconomic subgroups. We first identified subgroups where a developed AI tool underperformed by measuring algorithmic bias, where subgroups with depression were incorrectly predicted to be at lower risk than healthier subgroups. We then found inconsistencies between sensed-behaviors predictive of depression across these subgroups. Our findings suggest that researchers developing AI tools predicting mental health from sensed-behaviors should think critically about the generalizability of these tools, and consider tailored solutions for targeted populations.

Measuring algorithmic bias to analyze the reliability of AI tools that predict depression risk using smartphone sensed-behavioral data

AI tools intend to transform mental healthcare by providing remote estimates of depression risk using behavioral data collected by sensors embedded in smartphones. While these tools accurately predict elevated symptoms in small, homogenous populations, recent studies show that these tools are less accurate in larger, more diverse populations. In this work, we show that accuracy is reduced because sensed-behaviors are unreliable predictors of depression across individuals; specifically the sensed-behaviors that predict depression risk are inconsistent across demographic and socioeconomic subgroups. We first identified subgroups where a developed AI tool underperformed by measuring algorithmic bias, where subgroups with depression were incorrectly predicted to be at lower risk than healthier subgroups. We then found inconsistencies between sensed-behaviors predictive of depression across these subgroups. Our findings suggest that researchers developing AI tools predicting mental health from behavior should think critically about the generalizability of these tools, and consider tailored solutions for targeted populations.

From smartphone data to clinically relevant predictions: A systematic review of digital phenotyping methods in depression

BACKGROUND

Smartphone-based digital phenotyping enables potentially clinically relevant information to be collected as individuals go about their day. This could improve monitoring and interventions for people with Major Depressive Disorder (MDD). The aim of this systematic review was to investigate current digital phenotyping features and methods used in MDD.

METHODS

We searched PubMed, PsycINFO, Embase, Scopus and Web of Science (10/11/2023) for articles including: (1) MDD population, (2) smartphone-based features, (3) validated ratings. Risk of bias was assessed using several sources. Studies were compared within analysis goals (correlating features with depression, predicting symptom severity, diagnosis, mood state/episode, other). Twenty-four studies (9801 participants) were included.

RESULTS

Studies achieved moderate performance. Common themes included challenges from complex and missing data (leading to a risk of bias), and a lack of external validation.

DISCUSSION

Studies made progress towards relating digital phenotypes to clinical variables, often focusing on time-averaged features. Methods investigating temporal dynamics more directly may be beneficial for patient monitoring. European Research Council consolidator grant: 101001118, Prospero: CRD42022346264, Open Science Framework: https://osf.io/s7ay4.

Differential temporal utility of passively sensed smartphone features for depression and anxiety symptom prediction: a longitudinal cohort study

While studies show links between smartphone data and affective symptoms, we lack clarity on the temporal scale, specificity (e.g., to depression vs. anxiety), and person-specific (vs. group-level) nature of these associations. We conducted a large-scale (n = 1013) smartphone-based passive sensing study to identify within- and between-person digital markers of depression and anxiety symptoms over time. Participants (74.6% female; M age = 40.9) downloaded the LifeSense app, which facilitated continuous passive data collection (e.g., GPS, app and device use, communication) across 16 weeks. Hierarchical linear regression models tested the within- and between-person associations of 2-week windows of passively sensed data with depression (PHQ-8) or generalized anxiety (GAD-7). We used a shifting window to understand the time scale at which sensed features relate to mental health symptoms, predicting symptoms 2 weeks in the future (distal prediction), 1 week in the future (medial prediction), and 0 weeks in the future (proximal prediction). Spending more time at home relative to one's average was an early signal of PHQ-8 severity (distal β = 0.219, p = 0.012) and continued to relate to PHQ-8 at medial (β = 0.198, p = 0.022) and proximal (β = 0.183, p = 0.045) windows. In contrast, circadian movement was proximally related to (β = -0.131, p = 0.035) but did not predict (distal β = 0.034, p = 0.577; medial β = -0.089, p = 0.138) PHQ-8. Distinct communication features (i.e., call/text or app-based messaging) related to PHQ-8 and GAD-7. Findings have implications for identifying novel treatment targets, personalizing digital mental health interventions, and enhancing traditional patient-provider interactions. Certain features (e.g., circadian movement) may represent correlates but not true prospective indicators of affective symptoms. Conversely, other features like home duration may be such early signals of intra-individual symptom change, indicating the potential utility of prophylactic intervention (e.g., behavioral activation) in response to person-specific increases in these signals.

Towards Personalised Mood Prediction and Explanation for Depression from Biophysical Data

Digital health applications using Artificial Intelligence (AI) are a promising opportunity to address the widening gap between available resources and mental health needs globally. Increasingly, passively acquired data from wearables are augmented with carefully selected active data from depressed individuals to develop Machine Learning (ML) models of depression based on mood scores. However, most ML models are black box in nature, and hence the outputs are not explainable. Depression is also multimodal, and the reasons for depression may vary significantly between individuals. Explainable and personalised models will thus be beneficial to clinicians to determine the main features that lead to a decline in the mood state of a depressed individual, thus enabling suitable personalised therapy. This is currently lacking. Therefore, this study presents a methodology for developing personalised and accurate Deep Learning (DL)-based predictive mood models for depression, along with novel methods for identifying the key facets that lead to the exacerbation of depressive symptoms. We illustrate our approach by using an existing multimodal dataset containing longitudinal Ecological Momentary Assessments of depression, lifestyle data from wearables and neurocognitive assessments for 14 mild to moderately depressed participants over one month. We develop classification- and regression-based DL models to predict participants' current mood scores-a discrete score given to a participant based on the severity of their depressive symptoms. The models are trained inside eight different evolutionary-algorithm-based optimisation schemes that optimise the model parameters for a maximum predictive performance. A five-fold cross-validation scheme is used to verify the DL model's predictive performance against 10 classical ML-based models, with a model error as low as 6% for some participants. We use the best model from the optimisation process to extract indicators, using SHAP, ALE and Anchors from explainable AI literature to explain why certain predictions are made and how they affect mood. These feature insights can assist health professionals in incorporating personalised interventions into a depressed individual's treatment regimen.

Specific associations of passively sensed smartphone data with future symptoms of avoidance, fear, and physiological distress in social anxiety

Background

Prior literature links passively sensed information about a person's location, movement, and communication with social anxiety. These findings hold promise for identifying novel treatment targets, informing clinical care, and personalizing digital mental health interventions. However, social anxiety symptoms are heterogeneous; to identify more precise targets and tailor treatments, there is a need for personal sensing studies aimed at understanding differential predictors of the distinct subdomains of social anxiety. Our objective was to conduct a large-scale smartphone-based sensing study of fear, avoidance, and physiological symptoms in the context of trait social anxiety over time.

Methods

Participants (n = 1013; 74.6 % female; M age = 40.9) downloaded the LifeSense app, which collected continuous passive data (e.g., GPS, communication, app and device use) over 16 weeks. We tested a series of multilevel linear regression models to understand within- and between-person associations of 2-week windows of passively sensed smartphone data with fear, avoidance, and physiological distress on the self-reported Social Phobia Inventory (SPIN). A shifting sensor lag was applied to examine how smartphone features related to SPIN subdomains 2 weeks in the future (distal prediction), 1 week in the future (medial prediction), and 0 weeks in the future (proximal prediction).

Results

A decrease in time visiting novel places was a strong between-person predictor of social avoidance over time (distal β = -0.886, p = .002; medial β = -0.647, p = .029; proximal β = -0.818, p = .007). Reductions in call- and text-based communications were associated with social avoidance at both the between- (distal β = -0.882, p = .002; medial β = -0.932, p = .001; proximal β = -0.918, p = .001) and within- (distal β = -0.191, p = .046; medial β = -0.213, p = .028) person levels, as well as between-person fear of social situations (distal β = -0.860, p < .001; medial β = -0.892, p < .001; proximal β = -0.886, p < .001) over time. There were fewer significant associations of sensed data with physiological distress. Across the three subscales, smartphone data explained 9-12 % of the variance in social anxiety.

Conclusion

Findings have implications for understanding how social anxiety manifests in daily life, and for personalizing treatments. For example, a signal that someone is likely to begin avoiding social situations may suggest a need for alternative types of exposure-based interventions compared to a signal that someone is likely to begin experiencing increased physiological distress. Our results suggest that as a prophylactic means of targeting social avoidance, it may be helpful to deploy interventions involving social exposures in response to decreases in time spent visiting novel places.

Personality Science in the Digital Age: The Promises and Challenges of Psychological Targeting for Personalized Behavior-Change Interventions at Scale

With the rapidly growing availability of scalable psychological assessments, personality science holds great promise for the scientific study and applied use of customized behavior-change interventions. To facilitate this development, we propose a classification system that divides psychological targeting into two approaches that differ in the process by which interventions are designed: audience-to-content matching or content-to-audience matching. This system is both integrative and generative: It allows us to (a) integrate existing research on personalized interventions from different psychological subdisciplines (e.g., political, educational, organizational, consumer, and clinical and health psychology) and to (b) articulate open questions that generate promising new avenues for future research. Our objective is to infuse personality science into intervention research and encourage cross-disciplinary collaborations within and outside of psychology. To ensure the development of personality-customized interventions aligns with the broader interests of individuals (and society at large), we also address important ethical considerations for the use of psychological targeting (e.g., privacy, self-determination, and equity) and offer concrete guidelines for researchers and practitioners.

A Systematic Review of Location Data for Depression Prediction

Depression contributes to a wide range of maladjustment problems. With the development of technology, objective measurement for behavior and functional indicators of depression has become possible through the passive sensing technology of digital devices. Focusing on location data, we systematically reviewed the relationship between depression and location data. We searched Scopus, PubMed, and Web of Science databases by combining terms related to passive sensing and location data with depression. Thirty-one studies were included in this review. Location data demonstrated promising predictive power for depression. Studies examining the relationship between individual location data variables and depression, homestay, entropy, and the normalized entropy variable of entropy dimension showed the most consistent and significant correlations. Furthermore, variables of distance, irregularity, and location showed significant associations in some studies. However, semantic location showed inconsistent results. This suggests that the process of geographical movement is more related to mood changes than to semantic location. Future research must converge across studies on location-data measurement methods.

Personalised depression forecasting using mobile sensor data and ecological momentary assessment

Introduction

Digital health interventions are an effective way to treat depression, but it is still largely unclear how patients' individual symptoms evolve dynamically during such treatments. Data-driven forecasts of depressive symptoms would allow to greatly improve the personalisation of treatments. In current forecasting approaches, models are often trained on an entire population, resulting in a general model that works overall, but does not translate well to each individual in clinically heterogeneous, real-world populations. Model fairness across patient subgroups is also frequently overlooked. Personalised models tailored to the individual patient may therefore be promising.

Methods

We investigate different personalisation strategies using transfer learning, subgroup models, as well as subject-dependent standardisation on a newly-collected, longitudinal dataset of depression patients undergoing treatment with a digital intervention ( N = 65 patients recruited). Both passive mobile sensor data as well as ecological momentary assessments were available for modelling. We evaluated the models' ability to predict symptoms of depression (Patient Health Questionnaire-2; PHQ-2) at the end of each day, and to forecast symptoms of the next day.

Results

In our experiments, we achieve a best mean-absolute-error (MAE) of 0.801 (25% improvement) for predicting PHQ-2 values at the end of the day with subject-dependent standardisation compared to a non-personalised baseline ( MAE = 1.062 ). For one day ahead-forecasting, we can improve the baseline of 1.539 by 12 % to a MAE of 1.349 using a transfer learning approach with shared common layers. In addition, personalisation leads to fairer models at group-level.

Discussion

Our results suggest that personalisation using subject-dependent standardisation and transfer learning can improve predictions and forecasts, respectively, of depressive symptoms in participants of a digital depression intervention. We discuss technical and clinical limitations of this approach, avenues for future investigations, and how personalised machine learning architectures may be implemented to improve existing digital interventions for depression.

Day-to-day intrapersonal variability in mobility patterns and association with perceived stress: A cross-sectional study using GPS from 122 individuals in three European cities

Many aspects of our life are related to our mobility patterns and individuals can exhibit strong tendencies towards routine in their daily lives. Intrapersonal day-to-day variability in mobility patterns has been associated with mental health outcomes. The study aims were: (a) calculate intrapersonal day-to-day variability in mobility metrics for three cities; (b) explore interpersonal variability in mobility metrics by sex, season and city, and (c) describe intrapersonal variability in mobility and their association with perceived stress. Data came from the Physical Activity through Sustainable Transport Approaches (PASTA) project, 122 eligible adults wore location measurement devices over 7-consecutive days, on three occasions during 2015 (Antwerp: 41, Barcelona: 41, London: 40). Participants completed the Short Form Perceived Stress Scale (PSS-4). Day-to-day variability in mobility was explored via six mobility metrics using distance of GPS point from home (meters:m), distance travelled between consecutive GPS points (m) and energy expenditure (metabolic equivalents:METs) of each GPS point collected (n = 3,372,919). A Kruskal-Wallis H test determined whether the median daily mobility metrics differed by city, sex and season. Variance in correlation quantified day-to-day intrapersonal variability in mobility. Levene's tests or Kruskal-Wallis tests were applied to assess intrapersonal variability in mobility and perceived stress. There were differences in daily distance travelled, maximum distance from home and METS between individuals by sex, season and, for proportion of time at home also, by city. Intrapersonal variability across all mobility metrics were highly correlated; individuals had daily routines and largely stuck to them. We did not observe any association between stress and mobility. Individuals are habitual in their daily mobility patterns. This is useful for estimating environmental exposures and in fuelling simulation studies.

Depression Diagnosis and Forecast based on Mobile Phone Sensor Data

Previous studies have shown the correlation be-tween sensor data collected from mobile phones and human depression states. Compared to the traditional self-assessment questionnaires, the passive data collected from mobile phones is easier to access and less time-consuming. In particular, passive mobile phone data can be collected on a flexible time interval, thus detecting moment-by-moment psychological changes and helping achieve earlier interventions. Moreover, while previous studies mainly focused on depression diagnosis using mobile phone data, depression forecasting has not received sufficient attention. In this work, we extract four types of passive features from mobile phone data, including phone call, phone usage, user activity, and GPS features. We implement a long short-term memory (LSTM) network in a subject-independent 10-fold cross-validation setup to model both a diagnostic and a forecasting tasks. Experimental results show that the forecasting task achieves comparable results with the diagnostic task, which indicates the possibility of forecasting depression from mobile phone sensor data. Our model achieves an accuracy of 77.0 % for major depression forecasting (binary), an accuracy of 53.7 % for depression severity forecasting (5 classes), and a best RMSE score of 4.094 (PHQ-9, range from 0 to 27).

Opportunities for Smartphone Sensing in E-Health Research: A Narrative Review

Recent years have seen significant advances in the sensing capabilities of smartphones, enabling them to collect rich contextual information such as location, device usage, and human activity at a given point in time. Combined with widespread user adoption and the ability to gather user data remotely, smartphone-based sensing has become an appealing choice for health research. Numerous studies over the years have demonstrated the promise of using smartphone-based sensing to monitor a range of health conditions, particularly mental health conditions. However, as research is progressing to develop the predictive capabilities of smartphones, it becomes even more crucial to fully understand the capabilities and limitations of using this technology, given its potential impact on human health. To this end, this paper presents a narrative review of smartphone-sensing literature from the past 5 years, to highlight the opportunities and challenges of this approach in healthcare. It provides an overview of the type of health conditions studied, the types of data collected, tools used, and the challenges encountered in using smartphones for healthcare studies, which aims to serve as a guide for researchers wishing to embark on similar research in the future. Our findings highlight the predominance of mental health studies, discuss the opportunities of using standardized sensing approaches and machine-learning advancements, and present the trends of smartphone sensing in healthcare over the years.

Machine learning for passive mental health symptom prediction: Generalization across different longitudinal mobile sensing studies

Mobile sensing data processed using machine learning models can passively and remotely assess mental health symptoms from the context of patients' lives. Prior work has trained models using data from single longitudinal studies, collected from demographically homogeneous populations, over short time periods, using a single data collection platform or mobile application. The generalizability of model performance across studies has not been assessed. This study presents a first analysis to understand if models trained using combined longitudinal study data to predict mental health symptoms generalize across current publicly available data. We combined data from the CrossCheck (individuals living with schizophrenia) and StudentLife (university students) studies. In addition to assessing generalizability, we explored if personalizing models to align mobile sensing data, and oversampling less-represented severe symptoms, improved model performance. Leave-one-subject-out cross-validation (LOSO-CV) results were reported. Two symptoms (sleep quality and stress) had similar question-response structures across studies and were used as outcomes to explore cross-dataset prediction. Models trained with combined data were more likely to be predictive (significant improvement over predicting training data mean) than models trained with single-study data. Expected model performance improved if the distance between training and validation feature distributions decreased using combined versus single-study data. Personalization aligned each LOSO-CV participant with training data, but only improved predicting CrossCheck stress. Oversampling significantly improved severe symptom classification sensitivity and positive predictive value, but decreased model specificity. Taken together, these results show that machine learning models trained on combined longitudinal study data may generalize across heterogeneous datasets. We encourage researchers to disseminate collected de-identified mobile sensing and mental health symptom data, and further standardize data types collected across studies to enable better assessment of model generalizability.

A call for open data to develop mental health digital biomarkers

Digital biomarkers of mental health, created using data extracted from everyday technologies including smartphones, wearable devices, social media and computer interactions, have the opportunity to revolutionise mental health diagnosis and treatment by providing near-continuous unobtrusive and remote measures of behaviours associated with mental health symptoms. Machine learning models process data traces from these technologies to identify digital biomarkers. In this editorial, we caution clinicians against using digital biomarkers in practice until models are assessed for equitable predictions ('model equity') across demographically diverse patients at scale, behaviours over time, and data types extracted from different devices and platforms. We posit that it will be difficult for any individual clinic or large-scale study to assess and ensure model equity and alternatively call for the creation of a repository of open de-identified data for digital biomarker development.