Evaluating the Acceptability and Feasibility of Collecting Passive Smartphone Data to Estimate Psychological Functioning in U.S. Service Members and Veterans: A Pilot Study.

INTRODUCTION: This study investigated the acceptability and feasibility of digital phenotyping in a military sample with a history of traumatic brain injury and co-occurring psychological and cognitive symptoms. The first aim was to evaluate the acceptability of digital phenotyping by (1a) quantifying the proportion of participants willing to download the app and rates of dropout and app discontinuation and (1b) reviewing the stated reasons for both refusing and discontinuing use of the app. The second aim was to investigate technical feasibility by (2a) characterizing the amount and frequency of transferred data and (2b) documenting technical challenges. Exploratory aim 3 sought to leverage data on phone and keyboard interactions to predict if a participant (a) is depressed and (b) has depression that improves over the course of the study. MATERIALS AND METHODS: A passive digital phenotyping app (Mindstrong Discovery) functioned in the background of the participants' smartphones and passively collected phone usage and typing kinematics data. RESULTS: Fifteen out of 16 participants (93.8%) consented to install the app on their personal smartphone devices. Four participants (26.7%) discontinued the use of the app partway through the study, primarily because of keyboard usability and technical issues. Fourteen out of 15 participants (93.3%) had at least one data transfer, and the median number of days with data was 40 out of a possible 57 days. The exploratory machine learning models predicting depression status and improvement in depression performed better than chance. CONCLUSIONS: The findings of this pilot study suggest that digital phenotyping is acceptable and feasible in a military sample and provides support for future larger investigations of this technology.

Associations of smartphone usage patterns with sleep and mental health symptoms in a clinical cohort receiving virtual behavioral medicine care: a retrospective study.

Study Objectives: We sought to develop behavioral sleep measures from passively sensed human-smartphone interactions and retrospectively evaluate their associations with sleep disturbance, anxiety, and depressive symptoms in a large cohort of real-world patients receiving virtual behavioral medicine care. Methods: Behavioral sleep measures from smartphone data were developed: daily longest period of smartphone inactivity (inferred sleep period [ISP]); 30-day expected period of inactivity (expected sleep period [ESP]); regularity of the daily ISP compared to the ESP (overlap percentage); and smartphone usage during inferred sleep (disruptions, wakefulness during sleep period). These measures were compared to symptoms of sleep disturbance, anxiety, and depression using linear mixed-effects modeling. More than 2300 patients receiving standard-of-care virtual mental healthcare across more than 111 000 days were retrospectively analyzed. Results: Mean ESP duration was 8.4 h (SD = 2.3), overlap percentage 75% (SD = 18%) and disrupted time windows 4.85 (SD = 3). There were significant associations between overlap percentage (p < 0.001) and disruptions (p < 0.001) with sleep disturbance symptoms after accounting for demographics. Overlap percentage and disruptions were similarly associated with anxiety and depression symptoms (all p < 0.001). Conclusions: Smartphone behavioral measures appear useful to longitudinally monitor sleep and benchmark depressive and anxiety symptoms in patients receiving virtual behavioral medicine care. Patterns consistent with better sleep practices (i.e. greater regularity of ISP, fewer disruptions) were associated with lower levels of reported sleep disturbances, anxiety, and depression.

A framework for precision "dosing" of mental healthcare services: algorithm development and clinical pilot.

BACKGROUND: One in five adults in the US experience mental illness and over half of these adults do not receive treatment. In addition to the access gap, few innovations have been reported for ensuring the right level of mental healthcare service is available at the right time for individual patients. METHODS: Historical observational clinical data was leveraged from a virtual healthcare system. We conceptualize mental healthcare services themselves as therapeutic interventions and develop a prototype computational framework to estimate their potential longitudinal impacts on depressive symptom severity, which is then used to assess new treatment schedules and delivered to clinicians via a dashboard. We operationally define this process as "session dosing": 497 patients who started treatment with severe symptoms of depression between November 2020 and October 2021 were used for modeling. Subsequently, 22 mental health providers participated in a 5-week clinical quality improvement (QI) pilot, where they utilized the prototype dashboard in treatment planning with 126 patients. RESULTS: The developed framework was able to resolve patient symptom fluctuations from their treatment schedules: 77% of the modeling dataset fit criteria for using the individual fits for subsequent clinical planning where five anecdotal profile types were identified that presented different clinical opportunities. Based on initial quality thresholds for model fits, 88% of those individuals were identified as adequate for session optimization planning using the developed dashboard, while 12% supported more thorough treatment planning (e.g. different treatment modalities). In the clinical pilot, 90% of clinicians reported using the dashboard a few times or more per member. Although most clinicians (67.5%) either rarely or never used the dashboard to change session types, numerous other discussions were enabled, and opportunities for automating session recommendations were identified. CONCLUSIONS: It is possible to model and identify the extent to which mental healthcare services can resolve depressive symptom severity fluctuations. Implementation of one such prototype framework in a real-world clinic represents an advancement in mental healthcare treatment planning; however, investigations to assess which clinical endpoints are impacted by this technology, and the best way to incorporate such frameworks into clinical workflows, are needed and are actively being pursued.

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.

Combining digital pill and smartphone data to quantify medication adherence in an observational psychiatric pilot study.

Digital medicine systems (DMSs) offer a potential solution to increase medication adherence, which is an important barrier to treatment of psychiatric disorders. In this pilot, we enrolled N = 24 individuals diagnosed with severe mental illness to use an FDA-approved DMS for 5 months. We also collected digital phenotyping smartphone data to study behavioral associations with medication adherence. Our results suggest it is feasible to use the system, and we identified longitudinal associations between adherence and some of the communication-based phenotyping features. Larger studies and a focus on data quality are important next steps for this work.

Phase 3b Multicenter, Prospective, Open-label Trial to Evaluate the Effects of a Digital Medicine System on Inpatient Psychiatric Hospitalization Rates for Adults With Schizophrenia.

Objective: Inpatient psychiatric admissions drive the financial burden of schizophrenia, and medication adherence remains challenging. We assessed whether aripiprazole tablets with sensor (AS; system includes ingestible event-marker sensor, wearable sensor patches, and smartphone application) could reduce psychiatric hospitalizations compared with oral standard-of-care (SOC) antipsychotics.Methods: This phase 3b, mirror-image clinical trial was conducted from April 29, 2019-August 11, 2020, in adults with schizophrenia with ≥ 1 hospitalization in the previous 48 months who had been prescribed oral SOC for the preceding 6 months (retrospective phase). All participants used AS for at least 3 months and up to 6 months. Primary endpoint was the inpatient psychiatric hospitalization rate in the modified intent-to-treat (mITT; n = 113) population during prospective months 1-3 versus retrospective phase. Proportion of days covered by medication was the secondary endpoint. Safety endpoints included adverse events related to the medication or patch and suicidality.Results: AS significantly reduced hospitalizations during prospective months 1-3 (-9.7%) and months 1-6 (-21.3% [P ≤ .001 for all comparisons]) in the mITT population versus the corresponding retrospective phase. AS use improved confirmed medication ingestion by 26.5 percentage points in prospective months 1-3 (P ≤ .001) and reduced PANSS scores. Patches were well-tolerated, and no participant reported changes in suicide risk.Conclusions: Compared with oral SOC, AS reduced inpatient psychiatric hospitalization rates for adults with mild-to-moderate schizophrenia. The AS system may aid medication ingestion and is associated with improvements in symptoms, potentially reducing acute-care needs among patients with schizophrenia.Trial Registration: ClinicalTrials.gov identifier: NCT03892889.

Characterization of activity behavior using a digital medicine system and comparison to medication ingestion in patients with serious mental illness.

Activity patterns can be important indicators in patients with serious mental illness. Here, we utilized an accelerometer and electrocardiogram incorporated within a digital medicine system, which also provides objective medication ingestion records, to explore markers of patient activity and investigate whether these markers of behavioral change are related to medication adherence. We developed an activity rhythm score to measure the consistency of step count patterns across the treatment regimen and explored the intensity of activity during active intervals. We then compared these activity features to ingestion behavior, both on a daily basis, using daily features and single-day ingestion behavior, and at the patient-level, using aggregate features and overall ingestion rates. Higher values of the single-day features for both the activity rhythm and activity intensity scores were associated with higher rates of ingestion on the following day. Patients with a mean activity rhythm score greater than the patient-level median were also shown to have higher overall ingestion rates than patients with lower activity rhythm scores (p = 0.004). These initial insights demonstrate the ability of digital medicine to enable the development of digital behavioral markers that can be compared to previously unavailable objective ingestion information to improve medication adherence.

A Rest Quality Metric Using a Cluster-Based Analysis of Accelerometer Data and Correlation With Digital Medicine Ingestion Data: Algorithm Development.

BACKGROUND: Adherence to medication regimens and patient rest are two important factors in the well-being of patients with serious mental illness. Both of these behaviors are traditionally difficult to record objectively in unsupervised populations. OBJECTIVE: A digital medicine system that provides objective time-stamped medication ingestion records was used by patients with serious mental illness. Accelerometer data from the digital medicine system was used to assess rest quality and thus allow for investigation into correlations between rest and medication ingestion. METHODS: Longest daily rest periods were identified and then evaluated using a k-means clustering algorithm and distance metric to quantify the relative quality of patient rest during these periods. This accelerometer-derived quality-of-rest metric, along with other accepted metrics of rest quality, such as duration and start time of the longest rest periods, was compared to the objective medication ingestion records. Overall medication adherence classification based on rest features was not performed due to a lack of patients with poor adherence in the sample population. RESULTS: Explorations of the relationship between these rest metrics and ingestion did seem to indicate that patients with poor adherence experienced relatively low quality of rest; however, patients with better adherence did not necessarily exhibit consistent rest quality. This sample did not contain sufficient patients with poor adherence to draw more robust correlations between rest quality and ingestion behavior. The correlation of temporal outliers in these rest metrics with daily outliers in ingestion time was also explored. CONCLUSIONS: This result demonstrates the ability of digital medicine systems to quantify patient rest quality, providing a framework for further work to expand the participant population, compare these rest metrics to gold-standard sleep measurements, and correlate these digital medicine biomarkers with objective medication ingestion data.

Hummingbird Study: Results from an Exploratory Trial Assessing the Performance and Acceptance of a Digital Medicine System in Adults with Schizophrenia, Schizoaffective Disorder, or First-Episode Psychosis.

PURPOSE: Symptoms of psychotic disorders can complicate efforts to accurately evaluate patients' medication ingestion. The digital medicine system (DMS), composed of antipsychotic medication co-encapsulated with an ingestible sensor, wearable sensor patches, and a smartphone application, was developed to objectively measure medication ingestion. We assessed performance and acceptance of the DMS in subjects with psychotic disorders. METHODS: This was an 8-week open-label, single-arm, multicenter, Phase 4 pragmatic study (NCT03568500; EudraCT #2017-004602-17). Eligible adults were diagnosed with schizophrenia, schizoaffective disorder, or first-episode psychosis; were receiving aripiprazole, quetiapine, olanzapine, or risperidone; and could use the DMS with the application downloaded on a personal smartphone. The primary endpoint was good patch coverage, defined as the proportion of days over the assessment period where ≥80.0% of patch data was available, or an ingestion was detected. Exploratory endpoints included a survey on user satisfaction, used to assess acceptance of the DMS. Safety analyses included the incidence of treatment-emergent adverse events (TEAEs). RESULTS: From May 25, 2018 to March 22, 2019, 55 subjects were screened and 44 were enrolled. Good patch coverage was achieved on 63.4% of days assessed and the DMS generated an adherence metric of ≥80.0%, reflecting the percentage of ingestion events expected when good patch coverage was reported. Most subjects (53.5%) were satisfied with the DMS. Medical device skin irritations were the only TEAEs reported. CONCLUSION: The DMS had sufficient performance in, and acceptance from, subjects with psychotic disorders and was generally well tolerated.

Detection of Behavioral Anomalies in Medication Adherence Patterns Among Patients With Serious Mental Illness Engaged With a Digital Medicine System.

BACKGROUND: Adherence to medication is often represented in the form of a success percentage over a period of time. Although noticeable changes to aggregate adherence levels may be indicative of unstable medication behavior, a lack of noticeable changes in aggregate levels over time does not necessarily indicate stability. The ability to detect developing changes in medication-taking behavior under such conditions in real time would allow patients and care teams to make more timely and informed decisions. OBJECTIVE: This study aims to develop a method capable of identifying shifts in behavioral (medication) patterns at the individual level and subsequently assess the presence of such shifts in retrospective clinical trial data from patients with serious mental illness. METHODS: We defined the term adherence volatility as "the degree to which medication ingestion behavior fits expected behavior based on historically observed data" and defined a contextual anomaly system around this concept, leveraging the empirical entropy rate of a stochastic process as the basis for formulating anomaly detection. For the presented methodology, each patient's evolving behavior is used to dynamically construct the expectation bounds for each future interval, eliminating the need to rely on model training or a static reference sequence. RESULTS: Simulations demonstrated that the presented methodology identifies anomalous behavior patterns even when aggregate adherence levels remain constant and highlight the temporal dependence inherent in these anomalies. Although a given sequence of events may present as anomalous during one period, that sequence should subsequently contribute to future expectations and may not be considered anomalous at a later period-this feature was demonstrated in retrospective clinical trial data. In the same clinical trial data, anomalous behavioral shifts were identified at both high- and low-adherence levels and were spread across the whole treatment regimen, with 77.1% (81/105) of the population demonstrating at least one behavioral anomaly at some point in their treatment. CONCLUSIONS: Digital medicine systems offer new opportunities to inform treatment decisions and provide complementary information about medication adherence. This paper introduces the concept of adherence volatility and develops a new type of contextual anomaly detection, which does not require an a priori definition of normal and allows expectations to evolve with shifting behavior, removing the need to rely on training data or static reference sequences. Retrospective analysis from clinical trial data highlights that such an approach could provide new opportunities to meaningfully engage patients about potential shifts in their ingestion behavior; however, this framework is not intended to replace clinical judgment, rather to highlight elements of data that warrant attention. The evidence provided here identifies new areas for research and seems to justify additional explorations in this area.

Evaluating digital medicine ingestion data from seriously mentally ill patients with a Bayesian Hybrid Model.

The objective of this work was to adapt and evaluate the performance of a Bayesian hybrid model to characterize objective temporal medication ingestion parameters from two clinical studies in patients with serious mental illness (SMI) receiving treatment with a digital medicine system. This system provides a signal from an ingested sensor contained in the dosage form to a patient-worn patch and transmits this signal via the patient's mobile device. A previously developed hybrid Markov-von Mises model was used to obtain maximum-likelihood estimates for medication ingestion behavior parameters for individual patients. The individual parameter estimates were modeled to obtain distribution parameters of priors implemented in a Markov chain-Monte Carlo framework. Clinical and demographic covariates associated with model ingestion parameters were also assessed. We obtained individual estimates of overall observed ingestion percent (median:75.9%, range:18.2-98.3%, IQR:32.9%), rate of excess dosing events (median:0%, range:0-14.3%, IQR:3.0%) and observed ingestion duration. The modeling also provided estimates of the Markov-dependence probabilities of dosing success following a dosing success or failure. The ingestion-timing deviations were modeled with the von Mises distribution. A subset of 17 patients (22.1%) were identified as prompt correctors based on Markov-dependence probability of a dosing failure followed by a dosing success of unity. The prompt corrector sub-group had a better overall digital medicine ingestion parameter profile compared to those who were not prompt correctors. Our results demonstrate the potential utility of a Bayesian Hybrid Markov-von Mises model for characterizing digital medicine ingestion patterns in patients with SMI.

Hummingbird Study: a study protocol for a multicentre exploratory trial to assess the acceptance and performance of a digital medicine system in adults with schizophrenia, schizoaffective disorder or first-episode psychosis.

INTRODUCTION: In patients with schizophrenia, medication adherence is important for relapse prevention, and effective adherence monitoring is essential for treatment planning. A digital medicine system (DMS) has been developed to objectively monitor patient adherence and support clinical decision making regarding treatment choices. This study assesses the acceptance and performance of the DMS in adults with schizophrenia, schizoaffective disorder or first-episode psychosis and in healthcare professionals (HCPs). METHODS/ANALYSIS: This is a multicentre, 8-week, single-arm, open-label pragmatic trial designed using coproduction methodology. The study will be conducted at five National Health Service Foundation Trusts in the UK. Patients 18-65 years old with a diagnosis of schizophrenia, schizoaffective disorder or first-episode psychosis will be eligible. HCPs (psychiatrists, care coordinators, nurses, pharmacists), researchers, information governance personnel, clinical commissioning groups and patients participated in the study design and coproduction. Intervention employed will be the DMS, an integrated system comprising an oral sensor tablet coencapsulated with an antipsychotic, non-medicated wearable patch, mobile application (app) and web-based dashboard. The coencapsulation product contains aripiprazole, olanzapine, quetiapine or risperidone, as prescribed by the HCP, with a miniature ingestible event marker (IEM) in tablet. On ingestion, the IEM transmits a signal to the patch, which collects ingestion and physical activity data for processing on the patient's smartphone or tablet before transmission to a cloud-based server for viewing by patients, caregivers and HCPs on secure web portals or mobile apps. ETHICS AND DISSEMINATION: Approval was granted by London - City and East Research Ethics Committee (REC ref no 18/LO/0128), and clinical trial authorisation was provided by the Medicines and Healthcare products Regulatory Agency. Written informed consent will be obtained from every participant. The trial will be compliant with the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use guidelines and the Declaration of Helsinki. TRIAL REGISTRATION NUMBER: NCT03568500; EudraCT2017-004602-17; Pre-results.

A pharmacometric case study regarding the sensitivity of structural model parameter estimation to error in patient reported dosing times.

Although there is a body of literature focused on minimizing the effect of dosing inaccuracies on pharmacokinetic (PK) parameter estimation, most of the work centers on missing doses. No attempt has been made to specifically characterize the effect of error in reported dosing times. Additionally, existing work has largely dealt with cases in which the compound of interest is dosed at an interval no less than its terminal half-life. This work provides a case study investigating how error in patient reported dosing times might affect the accuracy of structural model parameter estimation under sparse sampling conditions when the dosing interval is less than the terminal half-life of the compound, and the underlying kinetics are monoexponential. Additional effects due to noncompliance with dosing events are not explored and it is assumed that the structural model and reasonable initial estimates of the model parameters are known. Under the conditions of our simulations, with structural model CV % ranging from ~20 to 60 %, parameter estimation inaccuracy derived from error in reported dosing times was largely controlled around 10 % on average. Given that no observed dosing was included in the design and sparse sampling was utilized, we believe these error results represent a practical ceiling given the variability and parameter estimates for the one-compartment model. The findings suggest additional investigations may be of interest and are noteworthy given the inability of current PK software platforms to accommodate error in dosing times.

Development and application of an aggregate adherence metric derived from population pharmacokinetics to inform clinical trial enrichment.

Nonadherence to prescribed medication is a common barrier to effective treatment, and current options to determine adherence are limited. This study describes development of an aggregate adherence measure based on population pharmacokinetics (PK), and its comparison to a subjective questionnaire, the Morisky 8-item medication adherence scale (MMAS8), in a trial of psychiatric patients on stable doses of oral aripiprazole. A comprehensive model was first built using plasma drug concentration data from 24 clinical studies comprising 448 patients with over 13,500 observations. Application of this model to independent patient profiles for a given drug-dosing regimen were used to generate the primary aggregate adherence metric, a ratio of observed versus expected plasma exposures at steady-state. Although the metric is capable of comparing relative adherence across groups, simulations showed that the metric is not sufficiently sensitive as an individual diagnostic in all cases. There were no trends observed between results from calculated aggregate adherence metrics and total scores from MMAS8 in a single-visit clinical trial of 47 patients with bipolar 1 disorder or schizophrenia who were on stable doses of aripiprazole, although a strong association was observed for one MMAS8 question. The range of the metric calculated for patients was between 0.16 and 3.15. The described approach of a novel "reverse" application of population PK to quantify relative adherence with an aggregate measure may be influential for both clinical and pharmacometric communities.

Genetic factors associated with gemcitabine pharmacokinetics, disposition, and toxicity.

AIM: The goal of this work was to investigate the associations of genetic and environmental factors with gemcitabine disposition and toxicity from genomewide data using a novel information theoretic approach. METHODS: We utilized the information theoretic K-way interaction information (KWII) metric to detect gene-gene and gene-environment interactions associated with gemcitabine disposition and gemcitabine-induced neutropenia in genomic and clinical data from Japanese cancer patients. RESULTS: The information theoretic KWII analyses identified age and four genes - DMD, HEXDC, CNTN4, and ALOX5AP - to be associated with gemcitabine pharmacokinetics (PK). The rs4769060 single-nucleotide polymorphism in the ALOX5AP gene was associated with all PK parameters studied. For gemcitabine-induced neutropenia, multiple associations with long intergenic noncoding RNA regions were detected. Pathway analysis identified leukotriene and eoxin synthesis, platelet homeostasis, and L1CAM interactions as potential pathways associated with gemcitabine disposition. CONCLUSION: The KWII analyses detected novel associations with gemcitabine PK and toxicity. These results could be used to inform future investigations involving gemcitabine efficacy in clinical settings.

An information theory analysis of gene-environmental interactions in count/rate data.

OBJECTIVE: To develop and critically evaluate an information theory method for identifying gene-gene and gene-environment interactions in count and rate data. METHODS: The entropy-based metric k-way interaction information (KWII) was critically assessed for utility in detecting interactions with count data and over-dispersed count data in three simulation studies of increasing complexity and in datasets from animal models of depression and colitis. The results were compared to Poisson regression. The power and effect size dependence of the KWII for detecting interactions was also assessed. RESULTS: The KWII was capable of effectively identifying the genetic and environmental predictors and their interactions in all three simulated datasets. The results indicate that the KWII approach may produce more parsimonious results than regression. In a rat model of depression, we successfully identified a prominent gender effect as well as other published associations. Analysis of severity scores from an animal model of colitis identified markers from chromosome 3, as well as unique first- and second-order associations for the individual sections of the colon and cecum. CONCLUSIONS: The results demonstrate the utility and versatility of our entropy-based method for gene-environment interaction analysis of count and rate data with Poisson and over-dispersed distributions.