Dataset Information

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

ABSTRACT:

Background

Although effective mental health treatments exist, the ability to match individuals to optimal treatments is poor, and timely assessment of response is difficult. One reason for these challenges is the lack of objective measurement of psychiatric symptoms. Sensors and active tasks recorded by smartphones provide a low-burden, low-cost, and scalable way to capture real-world data from patients that could augment clinical decision-making and move the field of mental health closer to measurement-based care.

Objective

This study tests the feasibility of a fully remote study on individuals with self-reported depression using an Android-based smartphone app to collect subjective and objective measures associated with depression severity. The goals of this pilot study are to develop an engaging user interface for high task adherence through user-centered design; test the quality of collected data from passive sensors; start building clinically relevant behavioral measures (features) from passive sensors and active inputs; and preliminarily explore connections between these features and depression severity.

Methods

A total of 600 participants were asked to download the study app to join this fully remote, observational 12-week study. The app passively collected 20 sensor data streams (eg, ambient audio level, location, and inertial measurement units), and participants were asked to complete daily survey tasks, weekly voice diaries, and the clinically validated Patient Health Questionnaire (PHQ-9) self-survey. Pairwise correlations between derived behavioral features (eg, weekly minutes spent at home) and PHQ-9 were computed. Using these behavioral features, we also constructed an elastic net penalized multivariate logistic regression model predicting depressed versus nondepressed PHQ-9 scores (ie, dichotomized PHQ-9).

Results

A total of 415 individuals logged into the app. Over the course of the 12-week study, these participants completed 83.35% (4151/4980) of the PHQ-9s. Applying data sufficiency rules for minimally necessary daily and weekly data resulted in 3779 participant-weeks of data across 384 participants. Using a subset of 34 behavioral features, we found that 11 features showed a significant (P<.001 Benjamini-Hochberg adjusted) Spearman correlation with weekly PHQ-9, including voice diary-derived word sentiment and ambient audio levels. Restricting the data to those cases in which all 34 behavioral features were present, we had available 1013 participant-weeks from 186 participants. The logistic regression model predicting depression status resulted in a 10-fold cross-validated mean area under the curve of 0.656 (SD 0.079).

Conclusions

This study finds a strong proof of concept for the use of a smartphone-based assessment of depression outcomes. Behavioral features derived from passive sensors and active tasks show promising correlations with a validated clinical measure of depression (PHQ-9). Future work is needed to increase scale that may permit the construction of more complex (eg, nonlinear) predictive models and better handle data missingness.

SUBMITTER: Nickels S

PROVIDER: S-EPMC8386379 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundToday, college students are dealing with depression at some of the highest rates in decades. As the primary mental health service provider, university counseling centers are limited in their capacity and efficiency to provide mental health care due to time constraints and reliance on students' self-reports. A mobile behavioral-sensing platform may serve as a solution to enhance the efficiency and accessibility of university counseling services.ObjectiveThe main objectives of this study are to (1) understand the usefulness of a mobile sensing platform (ie, iSee) in improving counseling services and assisting students' self-management of their depression conditions, and (2) explore what types of behavioral targets (ie, meaningful information extracted from raw sensor data) and feedback to deliver from both clinician and students' perspectives.MethodsWe conducted semistructured interviews with 9 clinicians and 12 students with depression recruited from a counseling center at a large Midwestern university. The interviews were 40-50 minutes long and were audio recorded and transcribed. The interview data were analyzed using thematic analysis with an inductive approach. Clinician and student interviews were analyzed separately for comparison. The process of extracting themes involved iterative coding, memo writing, theme revisits, and refinement.ResultsFrom the clinician perspective, the mobile sensing platform helps to improve counseling service by providing objective evidence for clinicians and filling gaps in clinician-patient communication. Clinicians suggested providing students with their sensed behavioral targets organized around personalized goals. Clinicians also recommended delivering therapeutic feedback to students based on their sensed behavioral targets, including positive reinforcement, reflection reminders, and challenging negative thoughts. From the student perspective, the mobile sensing platform helps to ease continued self-tracking practices. Students expressed their need for integrated behavioral targets to understand correlations between behaviors and depression. They also pointed out that they would prefer to avoid seeing negative feedback.ConclusionsAlthough clinician and student participants shared views on the advantages of iSee in supporting university counseling, they had divergent opinions on the types of behavioral targets and feedback to be provided via iSee. This exploratory work gained initial insights into the design of a mobile sensing platform for depression management and informed a more conclusive research project for the future.

Project description:BackgroundAdherence to computerized cognitive behavioral therapy (cCBT) programs in real-world settings can be poor, and in the absence of therapist support, effects are modest and short term. Moreover, because cCBT systems tend toward limited support and thus low-intensity treatment, they are typically most appropriate for people experiencing mild to moderate mental health difficulties. Blended therapy, that is, combining direct therapist contact with cCBT or psychoeducational materials, has been identified as one possible approach to address these limitations and widen access to individual CBT for depression. Building on the initial success of blended therapy, we explore an integrated approach that seeks to seamlessly combine face-to-face contact, electronic contact, and between-session activities. Integration also considers how the technology can support therapists' workflow and integrate with broader health care systems. The ultimate aim is to provide a structure within which therapists can deliver high-intensity treatments, while also greatly reducing face-to-face contact.ObjectiveThe research aimed to explore patients' and therapists' views on using a system for the delivery of individual treatment for depression that integrates face-to-face therapist contact with access to online resources and with synchronous online therapy sessions that allow collaborative exercises, and to establish design requirements and thus key design considerations for integrated systems that more seamlessly combine different modes of communication.MethodsWe conducted a series of four user-centered design studies. This included four design workshops and seven prototype testing sessions with 18 people who had received CBT for depression in the past, and 11 qualitative interviews and three role-play sessions with 12 CBT therapists experienced in the treatment of depression. Studies took place between July and December 2017 in Bristol, United Kingdom.ResultsWorkshops and prototyping sessions with people who had received CBT identified three important requirements for integrated platforms delivering CBT therapy for depression as follows: (1) features that help to overcome depression-related barriers, (2) features that support engagement, and (3) features that reinforce learning and support the development of new skills. Research with therapists highlighted the importance of the therapist and client working together, the impact of technology on therapists' workflow and workload, challenges and opportunities related to the use of online resources, and the potential of technology to support patient engagement. We use these findings to inform 12 design considerations for developing integrated therapy systems.ConclusionsTo meet clients' and therapists' needs, integrated systems need to help retain the personal connection, support both therapist- and patient-led activities, and provide access to materials and the ability to monitor progress. However, developers of such systems should be mindful of their capacity to disrupt current work practices and increase therapists' workload. Future research should evaluate the impact of integrated systems on patients and therapists in a real-world context.

Project description:BackgroundHealth care systems are rapidly deploying digital tools for disease management; however, few studies have evaluated their usability by vulnerable populations. To understand the barriers to app usage among vulnerable populations, we employed user-centered design (UCD) methods in the development of a new text messaging app.ObjectiveThe study aimed to describe variations in patients' engagement in the app design process, focusing on limited health literacy (LHL), limited English proficiency (LEP), and limited digital literacy (LDL).MethodsWe conducted 20 in-depth semistructured interviews with primary care patients at a public health care system, used open-ended discussions and card sorting tasks to seek input about mobile phones and text messaging, and used open coding to categorize the patterns of mobile phone usage and to evaluate engagement in the card sorting process. We examined qualitative differences in engagement by examining the extensiveness of participant feedback on existing and novel text messaging content and calculated the proportion of patients providing extensive feedback on existing and novel content, overall and by health literacy, English proficiency, and digital literacy.ResultsThe average age of the 20 participants was 59 (SD 8) years; 13 (65%) were female, 18 (90%) were nonwhite, 16 (80%) had LHL, and 13 (65%) had LEP. All had depression, and 14 (70%) had diabetes. Most participants had smartphones (18/20, 90%) and regularly used text messaging (15/20, 75%), but 14 (70%) of them reported having difficulty texting because of inability to type, physical disability, and low literacy. We identified 10 participants as specifically having LDL; 7 of these participants had LEP, and all 10 had LHL. Half of the participants required a modification of the card sorting activity owing to not understanding it or not being able to read the cards in the allotted time. The proportion of participants who gave extensive feedback on existing content was lower in participants with limited versus adequate English proficiency (4/13, 30% vs 5/7, 71%), limited versus adequate health literacy (7/16, 44% vs 3/4, 75%), and limited versus adequate digital literacy (4/10, 40% vs 6/10, 60%); none of these differences were statistically significant. When examining the proportion of patients who gave extensive feedback for novel messaging content, those with LHL were less engaged than those with adequate health literacy (8/16, 50% vs 4/4, 100%); there were no statistical differences by any subgroup.ConclusionsDespite widespread mobile phone use, digital literacy barriers are common among vulnerable populations. Engagement in the card sorting activity varied among participants and appeared to be lower among those with LHL, LEP, and LDL. Researchers employing traditional UCD methods should routinely measure these communication domains among their end-user samples. Future work is needed to replicate our findings in larger samples, but augmentation of card sorting with direct observation and audiovisual cues may be more productive in eliciting feedback for those with communication barriers.

Project description:BackgroundFalls account for approximately 50% of infant injury hospitalizations, and caretaker behavior is central to preventing infant falls. Behavior theory-informed interventions for injury prevention have been suggested, but to date, few have been reported. The potential of using smartphones for injury prevention intervention delivery is also underexploited.ObjectiveThis study aims to develop a behavior theory- and evidence-based as well as user-centered digital intervention as a mobile app for parents to prevent infant falls following agile development practices.MethodsInfant falls while feeding was selected as the fall mechanism to demonstrate the approach being taken to develop this intervention. In phase 1, the Behaviour Change Wheel was used as a theoretical framework supported by a literature review to define intervention components that were then implemented as a mobile app. In phase 2, after the person-based approach, user testing through think-aloud interviews and comprehension assessments were used to refine the content and implementation of the intervention.ResultsThe target behaviors identified in phase 1 were adequate rest for the newborn's mother and safe feeding practices defined as prepare, position, and place. From behavioral determinants and the Behaviour Change Wheel, the behavior change functions selected to achieve these target behaviors were psychological capability, social opportunity, and reflective motivation. The selected behavior change techniques aligned with these functions were providing information on health consequences, using a credible source, instruction on performing each behavior, and social support. The defined intervention was implemented in a draft Android app. In phase 2, 4 rounds of user testing were required to achieve the predefined target comprehension level. The results from the think-aloud interviews were used to refine the intervention content and app features. Overall, the results from phase 2 revealed that users found the information provided to be helpful. Features such as self-tracking and inclusion of the social and environmental aspects of falls prevention were liked by the participants. Important feedback for the successful implementation of the digital intervention was also obtained from the user testing.ConclusionsTo our knowledge, this is the first study to apply the Behaviour Change Wheel to develop a digital intervention for child injury prevention. This study provides a detailed example of evidence-based development of a behavior theory-informed mobile intervention for injury prevention refined using the person-based approach.

Project description:BackgroundAccounting for how end users engage with technologies is imperative for designing an efficacious mobile behavioral intervention.ObjectiveThis mixed methods analysis examined the translational potential of user-centered design and basic behavioral science to inform the design of a new mobile intervention for obesity and binge eating.MethodsA total of 22 adults (7/22, 32% non-Hispanic White; 8/22, 36% male) with self-reported obesity and recurrent binge eating (≥12 episodes in 3 months) who were interested in losing weight and reducing binge eating completed a prototyping design activity over 1 week. Leveraging evidence from behavioral economics on choice architecture, participants chose treatment strategies from 20 options (aligned with treatment targets composing a theoretical model of the relation between binge eating and weight) to demonstrate which strategies and treatment targets are relevant to end users. The process by which participants selected and implemented strategies and their change in outcomes were analyzed.ResultsAlthough prompted to select one strategy, participants selected between 1 and 3 strategies, citing perceived achievability, helpfulness, or relevance as selection reasons. Over the week, all practiced a strategy at least once; 82% (18/22) struggled with implementation, and 23% (5/22) added a new strategy. Several themes emerged on successes and challenges with implementation, yielding design implications for supporting users in behavior change. In postexperiment reflections, 82% (18/22) indicated the strategy was helpful, and 86% (19/22) planned to continue use. One-week average within-subject changes in weight (-2.2 [SD -5.0] pounds) and binge eating (-1.6 [SD -1.8] episodes) indicated small clinical improvement.ConclusionsApplying user-centered design and basic behavioral science yielded design insights to incorporate personalization through user choice with guidance, which may enhance engagement with and potential efficacy of digital health interventions.

Project description:BACKGROUND:Adolescents diagnosed with persistent asthma commonly take less than 50% of their prescribed inhaled corticosteroids (ICS), placing them at risk for asthma-related morbidity. Adolescents' difficulties with adherence occur in the context of normative developmental changes (eg, increased responsibility for disease management) and rely upon still developing self-regulation and problem-solving skills that are integral for asthma self-management. We developed an adaptive mobile health system, Responsive Asthma Care for Teens (ReACT), that facilitates self-regulation and problem-solving skills during times when adolescents' objectively measured ICS adherence data indicate suboptimal rates of medication use. OBJECTIVE:The current paper describes our user-centered and evidence-based design process in developing ReACT. We explain how we leveraged a combination of individual interviews, national crowdsourced feedback, and an advisory board comprised of target users to develop the intervention content. METHODS:We developed ReACT over a 15-month period using one-on-one interviews with target ReACT users (n=20), national crowdsourcing (n=257), and an advisory board (n=4) to refine content. Participants included 13-17-year-olds with asthma and their caregivers. A total of 280 adolescents and their caregivers participated in at least one stage of ReACT development. RESULTS:Consistent with self-regulation theory, adolescents identified a variety of salient intrapersonal (eg, forgetfulness, mood) and external (eg, changes in routine) barriers to ICS use during individual interviews. Adolescents viewed the majority of ReACT intervention content (514/555 messages, 93%) favorably during the crowdsourcing phase, and the advisory board helped to refine the content that did not receive favorable feedback during crowdsourcing. Additionally, the advisory board provided suggestions for improving additional components of ReACT (eg, videos, message flow). CONCLUSIONS:ReACT involved stakeholders via qualitative approaches and crowdsourcing throughout the creation and refinement of intervention content. The feedback we received from participants largely supported ReACT's emphasis on providing adaptive and personalized intervention content to facilitate self-regulation and problem-solving skills, and the research team successfully completed the recommended refinements to the intervention content during the iterative development process.

Project description:BackgroundMobile device-based tools to help adolescents practice skills outside of cognitive behavioral therapy (CBT) sessions for treating an anxiety disorder may lead to greater treatment gains.ObjectiveThis study aimed to develop, design, and test the acceptability, learnability, heuristics, and usability of MindClimb, a smartphone-based app for adolescents with anxiety to use between CBT sessions to plan and complete exposure activities using skills (cognitive, relaxation, exposure practice, and reward) learned in treatment.MethodsThis 3-phase study took place from August 2015 to December 2018. In phase 1, the app was designed and developed in consultation with young people and CBT therapists to identify desired functions and content. Feedback was subjected to thematic analysis using a general inductive approach. In phase 2, we conducted 2 high-fidelity testing sessions using the think-aloud approach (acceptability, learnability, usability) and 10-item System Usability Scale with 10 adolescents receiving CBT. The high-fidelity MindClimb app was evaluated by 5 app developers based on Nielsen's usability heuristics and 5-point severity ranking scale. In phase 3, a total of 8 adolescents and 3 therapists assessed the usability of MindClimb during CBT sessions by recording the frequency of skills practice, use of MindClimb features, satisfaction with the app, and barriers and facilitators to app use during treatment.ResultsFeedback from phase 1 consultations indicated that the app should (1) be responsive to user needs and preferences, (2) be easy to use and navigate, (3) have relevant content to the practice of CBT for anxiety, and (4) be aesthetically appealing. Using this feedback as a guide, a fully functional app prototype for usability testing and heuristic evaluation was developed. In phase 2, think-aloud and usability data resulted in minor revisions to the app, including refinement of exposure activities. The average system usability score was 77 in both testing cycles, indicating acceptable usability. The heuristic evaluation by app developers identified only minor errors (eg, loading speed of app content, with a score of 1 on the severity ranking scale). In phase 3, adolescents considered app features for completing exposure (6.2/10) and relaxation (6.4/10) modestly helpful. Both adolescents (average score 11.3/15, SD 1.6) and therapists (average score 10.0/12, 2.6 SD) reported being satisfied with the app.ConclusionsThe user-centered approach to developing and testing MindClimb resulted in a mobile health app that can be used by adolescents during CBT for anxiety. Evaluation of the use of this app in a clinical practice setting demonstrated that adolescents and therapists generally felt it was helpful for CBT practice outside of therapy sessions. Implementation studies with larger youth samples are necessary to evaluate how to optimize the use of technology in clinical care and examine the impact of the app plus CBT on clinical care processes and patient outcomes.

Project description:BackgroundDrug therapy in pediatric patients is a complex process. Children are subject to continuous growth and variation in drug-metabolizing enzyme activity, requiring continuous adaption of dosages. In Germany, currently no publicly available database exists that provides evidence-based information on drug dosages in pediatrics. For local drug dosing support, a prototype database has been developed within the Children's Hospital, Erlangen. A user-centered development process was initiated to establish an online platform for evidence-based dosing recommendations, as well as pharmacological and pharmaceutical drug information in pediatrics.ObjectivesThe objectives of the study were to survey the demand for such a platform and to assess the usability of the different versions of the developed system.MethodsThe developed prototype was evaluated in a pluralistic walkthrough with prospective end users. After a redesign, the second prototype of the online platform underwent an online usability testing based on a tailored questionnaire and the System Usability Scale (SUS) (n = 12).ResultsEleven of 12 participants expressed a demand for an online platform for pediatric dosing recommendations. The majority of the participants requested the integration of extended features, such as drug-drug interaction alerts, or information on adverse effects, pharmacokinetics, and pharmacodynamics. Particularly noteworthy is the demand for an online calculator; 5 of a total of 15 participants explicitly requested a calculator for dosages (based on age, weight, body surface) and glomerular filtration rate. The usability of the second prototype was rated "good to excellent" with a median SUS of 81.25.ConclusionLocal domain experts demand an online platform for pediatric dosing recommendations. The application of the user-centered design approach enabled the development of a prototype suitable for practical use. Multiple additional required functionalities have been identified, whereby the importance of an online calculator for patient-individual dosing recommendations was particularly emphasized.