Collaborative Projects

mDOT works with seven collaborative projects. The goal of mDOT is to create a new capability for researchers so they can discover, optimize, and deploy temporally precise mHealth interventions in real-life. Such interventions will be individualized to the moment-to-moment biopsychosocial-environmental context of each individual to directly prevent, manage or treat medical conditions. Our three TR&D’s are conducting research and developing mHealth technology to address this vision, but they can succeed only by working with real-life data and by addressing real-life health research problems. Our TR&D’s will work with a diverse group of external investigators, who are collecting data and investigating specific health issues using mHealth. This will ensure that the technologies developed by our TR&D’s can solve real problems facing the health research community and ensure the usability of these technologies by investigators who are external to the mDOT investigating team, representing the broader health research community.

Projects:


CP1: Novel Use of mHealth Data to Identify States of Vulnerability and Receptivity to JITAIs

Collaborating Investigators: Inbal Nahum-Shani, University of Michigan; David Wetter, University of Utah

Funding Status: U01CA229437-01; NIH/NCI; 9/1/18 – 8/31/23

Significance: CP1 is analyzing dense longitudinal sensor and self-report data collected in 5 field studies (3 completed and 2 ongoing) of ~1,500 smokers attempting to quit. Specifically, the project is investigating how the temporal dynamics and interactions of emotion, self-regulatory capacity (SRC), context, and other factors can be used to detect states of vulnerability and receptivity to just-in-time interventions, both of which are considered dynamically evolving latent states. Self-reported and sensor-based measures will be used to identify empirically-based and theoretically-grounded features across multiple time scales that are most predictive of engagement (i.e., usage of self-regulatory activities) and smoking lapse. CP1 hypothesizes that current receptivity is best represented by high positive activating emotions (e.g., happy, grateful), low negative deactivating emotions (e.g., sad, boredom), low craving, high self-efficacy, high self-regulatory capacity, and low risk contexts (e.g., specific locations, such as home). CP1 also hypothesizes that current vulnerability is represented by current and/or recent high negative emotions, low positive emotions, high craving, low self-efficacy, low self-regulatory capacity, and risky context (e.g., seeing others smoking).

Additionally, CP1 aims to investigate how knowledge of these latent states can be used to optimize real-time engagement in self-regulatory activities by conducting a Micro-Randomized Trial (MRT) of 150 smokers attempting to quit. Utilizing a mobile smoking cessation app, the MRT will randomize each individual multiple times per day to either (a) no intervention prompt; (b) a prompt recommending engagement in brief (low effort) strategies; or (c) a prompt recommending a more effortful practice of self-regulation strategies. CP1 aims to investigate whether, what type, and under what conditions (e.g., current state of vulnerability and/or receptivity) a prompt to engage the individual in self-regulatory activities increases engagement, hence reduces vulnerability. CP1 will be the first to yield a comprehensive conceptual, technical, and empirical foundation to develop effective just-in-time adaptive interventions based on dynamic models of vulnerability and receptivity.

Approach & Push Pull Relationship: CP1 will closely collaborate with all three TR&D’s. From sensor and self-report data of 1500 daily smokers, CP1 is developing a novel biomarker of self-efficacy. Initial biomarker development will not account for uncertainty, but a push activity with Aim 1 of TR&D1 will utilize uncertainty modeling tools to represent and quantity uncertainty in the self-efficacy biomarker derived from sensor-data. After validation of the uncertainty model using the collected data, Aim 2 will work with CP1 to train deep feature learning methods on large quantity of available mHealth biomarker data of stress, craving, self-efficacy, location, and other contexts, to improve the accuracy of predictions of risk and receptivity that incorporate measures of uncertainty. Given predictors of risk and receptivity, CP1 will work with TR&D1 to develop composite scores of vulnerability and receptivity which are predictive of risk of smoking lapse and opportunity to engage in self-regulatory activities, respectively. Aim 2 of TR&D1 will work with CP1 to characterize the time-varying relationships between biomarkers for stress, craving, self-efficacy, location, and other contexts and the risk of lapse and receptivity to engage in self-regulatory activities. Subsequently, CP1 will utilize the refined composite scores from Aim 2 and temporal dynamics of risk drivers from Aim 3 in an MRT, which will provide opportunities for evaluation and iterative refinement of the methods developed in TR&D1.

CP1 will work with TR&D2 to optimize decision rules to learn what type and under what conditions (e.g., current state of vulnerability and/or receptivity) a prompt to engage the individual in self-regulatory activities reduces vulnerability. Here, the treatment actions are different types of prompts and the near time reward will be measures of engagement in the self-regulatory activity as well as vulnerability. The MRT will provide data for TR&D2 to develop useful simulation testbeds for early evaluations of the methods under Aims 1,2. CP1 is interested in conducting feasibility studies that involve a TR&D2 RL personalization algorithm (based on work under Aims 1,2) that uses the above decision rules as a warm start to personalize these decision rules as an individual experiences the prompts. These studies will be used to iteratively refine methods under Aims 1,2 and this collaboration will lay the foundation for successful future research projects.

TR&D3 Aim 1 will work with CP1 to develop, test, iteratively refine, and deploy computationally efficient biomarkers of stress, craving, and self-efficacy derived from wrist-worn PPG sensors that can run in real-time and be used in the MRT trial. High rate multispectral PPG and motion sensor architecture design from Aim 2 will enable reliable beat-to-beat heart rate estimation to support real-time computation of biomarkers. CP1 will test and provide iterative feedback for improvement of both the sensor and biomarkers computed in real-time.


CP2: Personalized Digital Behavior Change Interventions to Promote Oral Health

Collaborating Investigator: Vivek Shetty, University of California, Los Angeles

Funding Status: UG3DE028723; NIH/NIDCR/HHS; 04/01/19 - 03/31/25

Significance: CP2, a collaborative project involving industry (P&G, Delta Dental), will remotely monitor Oral Health Behaviors (OHBs) in the home setting and use the insights to develop a computationally-driven, personalized Digital Oral Health Intervention (DOHIs) and test its real-world efficacy in engaging at-risk individuals in ideal OHBs and improving their oral health. CP2 builds on the Remote Oral Behavior Assessment System project (ROBAS; 1R01DE025244; NIH/NIDCR; 7/1/15–5/31/20: PI: Shetty) that provides objective, individual-level and ecologically-valid data on oral hygiene behaviors.

In the preparatory UG3 phase, CP2 will engage end-users in the co-design of an oral self-care app and establish the usability and feasibility of the system. In the UH3 phase, CP2 will build and validate computational models for inferring the quality of OHBs and for tailoring of the DOHI. Using a cohort of 130 subjects, CP2 will conduct a 10-week Micro-Randomized Trial (MRT) to optimize the adaptive tailoring of the engagement strategies for the DOHI. Finally, CP2 will test its hypothesis (that the dynamic and personalized DOHI will be more effective than traditional, static, clinician-delivered OHI in improving oral health and adherence to 2x2x4 OHBs (brush 2 times a day, for 2 minutes each time, all 4 dental quadrants)) through a 6-month, pragmatic, randomized, controlled, parallel-group clinical trial of 260 subjects.

CP2 can fundamentally alter oral healthcare, emphasizing prevention and oral health maintenance. Beyond advancing behavior change theories, linking dental disease to actual OHBs would enhance our understanding of dental disease determinants and establish which digital behavioral intervention may be most effective, when and for whom, providing a springboard to the practice of 21st century temporally-precise dentistry.

Approach & Push Pull Relationship: CP2 is developing digital biomarkers of brushing from consumer toothbrushes and wearable wrist. To analyze the time series of OHB biomarkers from its studies, CP2 will leverage tools developed by TR&D1 Aim 1 to characterize the uncertainty in its biomarkers. In collaboration with TR&D1 Aim 2, CP2 will develop a dynamic risk indicator for dental disease which will support future intervention development. Successful data collection over six months requires enough engagement so participants can continue to be interested in the study. TR&D1 will work with CP2 to develop an score of engagement and TR&D2 will work with CP2 to identify strategies for improving engagement from the data collected so that future such studies can achieve greater compliance from their participants over long-term. Next, CP2 will utilize modeling tools from TR&D1 Aim 3 to model the dynamic relationships among the sociobehavioral risk factors captured by digital biomarkers of OHBs, eating, stress, activity, location, and mobility. Multiple iterations of model development, evaluation on study data, and refinement with domain experts will ensure that TR&D1 methods advance CP2 research on temporally-precise dentistry.

In push/pull collaboration with TR&D2, CP2 will develop DOHI intervention decision rules based on data from the 10-week Micro-Randomized Trial (MRT). In the RL framework, the actions are engagement strategies and the near time rewards is daily engagement in the 2x2x4 OHBs. The MRT will also provide data for TR&D2 to develop useful simulations for early evaluations of the methods under Aims 1 and 2. TR&D2 will provide a RL personalization algorithm (based on work under Aims 1 and 2) that uses the above decision rules as a warm start to personalize these decision rules as an individual experiences the DOHI intervention. CP2 is interested in deploying the developed RL algorithm in real-time to personalize decision rules for a small additional number of subjects in the 6-month trial; this deployment will provide data on real-time feasibility and acceptability from subjects as they experience the RL algorithm. The 6-month trial will also provide data to build a testbed simulation for use by TR&D2 to conduct early evaluation of the methods under Aim 3.

Currently, OHB biomarkers are being implemented in a cloud platform, after data collection. TR&D3 Aim 1 will work with CP2 to design, develop and validate micromarkers for real-time detection of OHB’s on smartphones from wrist and brush sensors. These new biomarkers will then become easily usable by researchers deploying wrist-worn sensors and in common consumer devices (e.g., Oral-B). In addition, integrating information across the sensors embedded into the toothbrush and wristband will provide unique insight on how the subjects hold and use their toothbrush. For this purpose, CP2 will test and deploy distributed fusion of information from two high rate inertial sensors at high time resolution from Aim 2. The UG3 will provide rapid feedback on the usability and utility of TR&D1 sensors and methods, which will be evaluated in the UH3 phase of CP2.


CP3: Operationalizing Behavioral Theory for mHealth: Dynamics, Context, and Personalization

Collaborating Investigator: Predrag Klasnja, University of Michigan

Funding Status: 1U01CA229445-01; NIH/NCI; 09/19/2018–08/31/2022

Significance: The long-term goal of CP3 is to enable the transformative potential of mHealth by addressing the behavior-theoretic, measurement, modeling, and intervention design challenges and opportunities presented by intensively collected longitudinal data. CP3 will investigate these issues by focusing on physical activity and sedentary behavior. To validate the proposed research, CP3 builds on the NIH-funded HeartSteps trial, which CP3 collaborator Klasnja leads. HeartSteps is a year-long micro-randomized trial (MRT) of an adaptive mHealth intervention based on Social-Cognitive Theory (SCT) that aims to increase walking and decrease sedentary behavior in a cohort of 60 patients with Stage 1 hypertension.

CP3 aims to develop and refine measures of theoretical constructs that influence behaviors and intervention response. Based on methods advanced in NIH’s Science of Behavior Change, CP3 will refine measures of dynamic theoretical constructs hypothesized by SCT to shape our target behaviors, as well as develop measures of constructs postulated by the Dual Process theories. Measures will be developed or refined to enable modeling of intensive longitudinal data about psychosocial and contextual influences on walking and sedentary behavior at different time scales, from hourly to monthly. Further, CP3 will enhance the existing HeartSteps trial with additional measures and recruit a second cohort of sedentary overweight/obese, but otherwise healthy adults. HeartSteps employs novel sources of information (e.g. wearable sensors, users’ calendars, location and other smartphone data) to obtain measures that were previously dependent on self-report. In this study, CP3 will enrich HeartSteps with the developed measures and add a second cohort of 60 sedentary overweight/obese, but otherwise healthy adults. The two HeartSteps cohorts will provide data needed to validate the proposed measures as well as to support model development and validation. Specifically, CP3 includes research on operationalizing dynamic and contextualized theories of behavior in naturalistic and interventional settings within the dynamic Bayesian network model framework, including learning personalized models and warm-starting personalization from population-level models.

Approach & Push Pull Relationship: Both TR&D1 and TR&D2 will work with CP3 to ensure that the methods developed are grounded in real-life needs and to ensure that the technologies developed are readily usable. The HeartSteps cohort data contains rich multimodal mHealth biomarker time series with complex patterns of noise and missingness (different from the case of oral health biomarkers of CP2). CP3 will benefit from uncertainty models from this iterative collaboration, while the size and complexity of the data will provide an opportunity for thorough empirical evaluation and validation of the TR&D1 Aim 1 approach. Another important issue for CP3 is the potentially high risk for participant disengagement given the one year duration of the HeartSteps study. CP3 will work with TR&D1 Aim 3 to develop novel composite scores of disengagement risk and receptivity to engagement interventions. This will provide an opportunity to TR&D1 to extend the methods of Aim 3 to a novel setting which differs significantly from the risk scores related to smoking lapse, dental disease, and other use cases. Temporal triggers and risk factors for disengagement and receptivity will be identified in an iterative process and compiled into the composite score. Disengagement outcomes from HeartSteps cohort will be used to refine and validate the resulting scores.

CP3 will collaborate with TR&D2 on all three specific aims. In particular, CP3 needs to account for delayed effects due to user habituation (Aim 1). CP3 will contribute data and collaborate on constructing the warm-start population-level baseline models for the personalization of decision rules, under Aim 2. This will push the boundaries of personalization methods beyond the via traditional (high variance, low bias) person-specific or (low variance, high bias) population-based algorithms. A fundamental challenge CP3 is confronting is that it utilizes interventions operating at different time scales and with different proximal outcomes. Currently, CP3 assumes that the decision rules for all of these interventions can be learned independently. However, CP3 recognizes that burden imposed by one type of intervention is likely to spill over and reduce effectiveness of interventions at other time scales. Thus, the work under Aim 3 by TR&D2 is critical to CP3. CP3 is committed to including both the methods for accommodating the delayed effects under Aim 1 as well as the personalization algorithm in their updated version of the HeartSteps application and conducting a feasibility study for use in informing future research directions of both CP3 and TR&D 2. CP3 will provide a real-life evaluation of the methods developed under all three specific aims of TR&D2, and contribute to iterative refinement, as participants experience these algorithms over the duration of one year in the study.


CP4: Control Systems Engineering for Counteracting Notification Fatigue: An Examination of Health Behavior Change

Collaborating Investigators: Daniel E. Rivera (PI), Arizona State University

Funding Status: R01LM013107; NIH/NLM; 03/06/19 - 02/28/23

Significance: A wide range of technologies associated with mHealth, such as smartphones, wearables (e.g., Fitbit, Apple Watch), and medical devices use alerts to inspire actions of users. Potentially useful alerts run a risk of alert fatigue, whereby individuals ignore alerts over time. For example, physical activity interventions use alerts to inspire activity. These notifications work initially, but their efficacy diminishes over time. CP4 seeks to develop novel solutions for reducing alert fatigue by applying advanced analytical procedures inspired by dynamical systems modeling and control systems engineering. Location, digital traces, and other digital data facilitate inferencing states when a person would desire or need alerts (which, in CP4 are referred to as “just-in-time” states). For example, a suggestion to walk (e.g., SMS saying, "Want to go for a walk?") may only produce the desired outcome when a person's state (e.g., low stress) and context (e.g., no meetings, nice weather) align such that the person appreciates the notification (“receptivity”) and can act on it (“opportunity”). Estimating the likelihood that a given moment is a just-in-time state requires not only data but also an approach to manage the multivariate, dynamic, idiosyncratic, and multi-timescale nature of the problem. Further, just-in-time notifications cannot be viewed in a vacuum and, instead, are often part of a more long-term process, such as sustained engagement in a health behavior, thus making it a multi-timescale problem. The purpose of this work is to develop a just-in-time state estimation strategy and to stage a multi-timescale controller for walking as a concrete use-case of a control systems approach to counteract alert fatigue.

Efforts in CP4 consist of using Model-on-Demand estimation (an instance-based machine learning methodology that integrates local and global modeling and has been widely adopted within the control engineering community) as the framework for a novel, just-in-time (JIT) state estimator applicable to mHealth. Its development will be initially tested on secondary data from the HeartSteps intervention v1 (described in CP3) that will be followed by an in-the-field validation experiment involving a micro-randomized trial. The validated JIT estimator will be linked with a daily timescale model created in previous work and a multi-timescale model predictive controller to support decision-making about when to send notifications and personalized, adaptive daily step goals to support accumulation of walking bouts into meaningful change. The final control system will be evaluated in a small cohort study.

Approach & Push Pull Relationship: CP4 will initially focus on the analysis of the HeartSteps v1 data set using a Model-on-Demand estimation procedure for inferring just-in-time states. The primary proximal outcomes in HeartSteps v1 is 30-minute step count, as recorded by a Jawbone fitness tracker. This device has the property that if the device is worn and no steps are detected, no step count is recorded. Thus, both a zero-step count and missing step count (e.g., due to lack of connectivity, device not being worn, device not being on, etc.) appear as missing data in the HeartSteps v1 data set. Initial treatment efficacy analysis performed using the HeartSteps v1 data set was based on zero imputation of missing step values, discarding the uncertainty due to missingness. The methods developed under TR&D1 Aim 1 will be applied to the HeartSteps v1 data set to model and represent the uncertainty due to missing step count information. While CP3 will test the clinical utility of uncertainty model in interventions in its MRT, CP4 will assess the utility of the TR&D1 in dynamical systems modeling over multiply imputed data sets. The two teams will work together to improve the missing step count models and analyze the impact of missing step count modeling on end-to-end inference of just-in-time states. The powerful modeling framework developed in CP4 will provide a rigorous and well-specified setting in which to create and refine uncertainty models through multiple push-pull iterations between the two teams.


CP5: Affective Science and Smoking Cessation: Real-time Real-world Assessment

Collaborating Investigators: Cho Lam (PI) & David Wetter, University of Utah

Funding Status: R01CA224537 ; NIH/NCI; 1/19/18 – 12/31/22

Significance: Smoking cessation is a cornerstone of cancer risk reduction, as 30% of all cancers are directly attributable to tobacco. Although half of all current smokers attempt to quit each year, over 90% of those quit attempts fail. Numerous conceptual models and robust empirical evidence suggests that affect is a potent determinant of smoking lapse. For example, higher lapse risk and lower rates of long-term abstinence are associated with acute experiences of negative emotions such as anger, sadness, and guilt, while acute experiences of positive emotions such as happiness, pride, and gratitude have the opposite effect. A key factor in abstinence is the self-regulatory capacity (SRC), which is an individual’s ability to exert control over behavior, thoughts, and emotions. It is based on the constellation and temporal dynamics of emotion, context, and other factors. Unfortunately, very little is known about the temporal dynamics and interactions of emotion, context, and SRC in real time under field conditions, as well as how these complex combinations influence lapse risk. This lack of knowledge severely hampers both our conceptual models and our ability to develop just-in-time, adaptive interventions (JITAIs) that effectively target the real-time, real-world self-regulatory mechanisms underlying successful cessation. The three goals of CP5 are to 1) leverage sensor-derived biomarkers to obtain objective real-time assessments of lapse, SRC, and geographic place (i.e. context); 2) integrate self-reports of distinct positive and negative emotions with sensor-derived biomarkers to model the time-varying dynamics of affect and the impact of context on affective experience; and 3) develop predictive models of lapse risk based on affect, SRC, and context which can form the basis for future JITAIs.

While CP1 is investigating methods to engage newly abstinent smokers in self-regulatory intervention using biomarkers of SRC, self-efficacy, stress, and other contexts, CP5 is analyzing the mobile sensor and self-report data collected in its observational study of smokers to develop biomarker of SRC and predictive models of lapse risk from mHealth biomarkers of affect, SRC, and other contexts.

Approach & Push Pull Relationship:  CP5 is recruiting daily smokers interested in quitting who wear the AutoSense chestband and MotionSense wrist sensors for one week prior to quitting and two weeks post-quit. They also carry a study-provided smartphone. From the sensor data, CP5 is obtaining sensor-derived biomarkers of stress, geoexposure, and smoking, and EMA’s provide additional assessment of emotions and context. CP5 will work with Aims 1, 2, and 3 of TR&D1. An initial model for SRC under development by CP5 does not explicitly account for uncertainty in sensor-derived biomarkers or in EMA-based assessments resulting from missingness. A push activity will utilize uncertainty models developed in Aim 1 of TR&D1 to represent and quantify uncertainty in sensor- and EMA-derived biomarkers. In a pull activity, we will jointly characterize the resulting uncertainty in the model for SRC. This will enable researchers to take measurement of uncertainty into account when analyzing the linkage between SRC and distinct emotions, EMA-derived context, sensor-derived states of stress and craving, and geographic context. In addition, uncertainty representation for the derived SRC biomarker can inform downstream analysis of the link between SRC and lapse risk. CP5 will work with TR&D1 Aim 2 to identify timing triggers from affect, context, SRC, and other lapse risks from the collected data, to guide the delivery of temporally-precise interventions. CP5 will utilize tools developed in Aim 2 for identifying specific temporal patterns that predict future lapse risk. Analysis of the effectiveness of these triggers on collected data will lead to a pull activity in which the identified temporal patterns are distilled into a composite risk score which can be computed from sensor data in real-time and characterizes the time-varying risk level. The resulting time triggers and risk scores will be utilized in the MRT which will be deployed by CP1. CP5 will utilize methods from TR&D1 Aim 3 for modeling the dynamic relationships between positive and negative discrete emotions, and biomarkers for stress, craving, and context that inform the temporal variations in SRC and lapse risk. Initially, the models will describe the lead-lag relationships between covariates. When sufficient sensor data has been collected, a pull activity with Aim 3 of TR&D1 will jointly develop deep feature learning methods based on variational autoencoders that can learn nonlinear temporal features to improve the accuracy in predicting the time-varying risk of lapse. This can help discover the optimal content and timing for temporally-precise interventions. The iterations between the two teams in testing the initial model, revising the model, and testing again will eventually lead to methods for all three aims that will be directly applicable to smoking cessation research using mHealth biomarkers.


CP6: Non-invasive Biosensors to Detect Cardiovascular Changes in Heart Failure

Collaborating Investigators: Omer Inan, Georgia Institute of Technology

Funding Status: 1R01HL130619-01A1; NIH/NHLBI; 04/01/2016 – 03/31/2021

Significance: CP6 is developing an unobtrusively wearable system for continuously monitoring heart failure (HF) patients in naturalistic settings to automatically assess their risk of experiencing an exacerbation so that appropriate and timely feedback can be provided to caregivers and to the patients themselves. Given the high cost and mortality associated with decompensation and associated rehospitalizations in heart failure, it is imperative that novel noninvasive methods be developed for assessing the risk of decompensation that can support interventions to prevent hospitalization. CP6 is employing ballistocardiography (BCG) to both characterize the state of compensation and decompensation over time in HF patients, and to measure hemodynamic responses to stressors experienced in normal activities of daily living (e.g., walking, climbing stairs). It is investigating the underlying mechanisms involved in the genesis of wearable BCG signals that can measure reaction forces propagating through the body in response to the heartbeat to determine the placement and modality of the sensors for optimizing the measurement of heart function.

CP6 has designed and implemented multiple sensing approaches to heart function assessment that can be used by patients in their homes, including a weighing scale with modified load cells that is capable of sensing displacements due to heart contractions, and prototype wearable sensors that can be worn on the torso of the patient and uses accelerometry to measure reaction forces. CP6 has conducted a study with HF patients that are admitted with decompensation and treated during hospitalization to restore compensation. BCG signals were collected from patients at intake and during hospitalization, and in outpatient follow-up at home. Analysis of these signals has demonstrated that there is significant differences in BCG variability between the decompensated and compensated HF states within the same patient. On-going work is assessing the role of individual differences in the characteristics of the BCG signal and is exploring the automated classification of states of decompensation from BCG signals. CP6 will conduct feasibility and usability testing to assess the feasibility and practicality of using BCG signals collected in the home setting to identify patients at risk for decompensation. An additional goal of CP6 is to develop novel machine learning algorithms to estimate cardiac output (CO), blood pressure (BP), and indirect calorimetry from ambulant subjects in a home setting using wearable BCG technology.

Approach & Push Pull Relationship: CP6 is generating novel sensor and biomarker models for estimating a decompensation score, along with cardiac output (CO), blood pressure (BP), and indirect calorimetry. Currently, high rate sensor measurements are collected in the field and for subsequent offline processing and the computation of biomarkers. In order to achieve its long-term goal of providing timely feedback to patients and caregivers, these biomarker models need to be implemented efficiently on the sensors and the computed biomarkers should be easily accessible on a smartphone in real-time. Such an implementation should be packaged in a smartphone app so it can provide repeatable measurements without consuming significant computation or battery capacity, thereby hiding the complexity of data screening, cleaning, processing and machine learning algorithms from researchers who are using it in their research studies. To transition these novel biomarkers into a mobile setting for real-time assessment, TR&D3 will collaborate closely with CP6 and undertake the following tasks: In alignment with TR&D3 Aim 1, 1) work with the CP6 technical team to identify micromarkers that can be efficiently computed at the sensor and can support biomarker inferences at the smartphone from micromarkers computed from the high rate sensor data 2.) Implement a computationally efficient and real-time version of biomarkers computed at the sensor in concert with a smartphone app for CP6 that they can test in their cohort of HF patients. With iterative feedback, the implementation of these real-time biomarkers can be improved so that they provide comparable sensitivity and specificity as the non-real-time version of the biomarkers originally developed by CP6.

CP6 will investigate the value of co-optimized hardware/software designs of high data rate sensors from TR&D3 Aim 2 in estimating timing events on the cardiac cycle captured by the mDOT sensors (RFPatch and MotionSenseHRV) in complementing the timing information provided by the ECG and Ballistography sensors of CP6. TR&D3 team will implement middleware services to support the distributed computation of the time interval between these cardiac events captured at different sensors and iterate this implementation to achieve the precision required by CP6 biomarker computations. Iterative testing and deployment of these sensors in CP6 participants will be used in improving the architectures and algorithms.


CP7: mProv: Provenance-based Data Analytics Cyberinfrastructure for High-frequency Mobile Sensor Data

Collaborating Investigator: Zach Ives, University of Pennsylvania (PI: Santosh Kumar, University of Memphis)

Funding Status: ACI-1640183; NSF/ACI; 09/01/16 - 08/31/21

Significance: There are major hurdles to using mobile sensor data to advance research on computational modeling of human health and behavior, including lack of access to high-quality mobile sensor data, regulatory obligations in accessing and using mobile sensor data collected from humans, and a lack of metadata capture and access services for the provenance, quality, and integrity of the data and inferences made from it. CP7 is developing a new cyberinfrastructure called mProv to annotate high-frequency mobile sensor data with data source, quality, validity, and semantics to facilitate the sharing of such data with the wider research community for third party research. It is developing techniques to integrate metadata and data capture over mobile streaming data, and propagate such data in order to enable reasoning about uncertainty and variability; runtime infrastructure and APIs for efficient sensor data acquisition and reply (integrated with human data capture), and mechanisms for managing privacy policies. To support interpretation of sensor-derived features and inferences (i.e., markers of health, behavior, and context) by researchers (for concurrent development that makes use of datastreams developed by other researchers) and automating analysis by machines, CP7 has developed datastream representation to support a common metadata structure that allows both mCerebrum and Cerebral Cortex (installed on mobile phones and the cloud respectively) to annotate the datastream with metadata. It has also developed storage, interface, instrumentation, and visualization tools for provenance tracking through stream processing operators. Provenance information can be automatically captured as a series of entities, activities, and relationships in a graph database from which it can be queried or visualized, even in near-real-time. It has built a core provenance repository with user authentication, group creation, and metadata storage capabilities. These are exposed through a simple REST microservices framework, and they can be retargeted at the back-end to a variety of SQL and NoSQL database systems. Currently, CP7 software uses Cassandra, REDIS, Neo4J, and Postgres. CP7 also works closely with an R24 from NIBIB (R24EB025845; PI: Ida Sim, UC San Francisco; 7/1/17-6/30/20) to standardize biomarkers that have been validated and being adopted in the research community via an IEEE Working Group (P1752, Open Mobile Health Standards). To evaluate its work under realistic settings, CP7 is conducting multiple iterations of 100-day field studies in 100 participants in collaboration with the Open Humans project to generate open data set that can be used by researchers to develop mHealth biomarkers for detecting daily stressors.

Approach & Push Pull Relationship: CP7 will get access to novel biomarkers from TR&D3 Aim 1 so that it can develop and implement appropriate annotations, provenance, and pursue standardization. This step will enable the adoption of these biomarkers by the wider research community, especially among researchers working on secondary analysis of existing datasets. A particularly relevant category of metadata that CP7 is concerned with is that relating to privacy. These include metadata describing context-dependent privacy policy that govern downstream sharing and use of sensor measurements and derivative micromarkers and biomarkers. It also includes sharing of metadata capturing transformation (e.g. sanitization, addition of noise etc.) that data stream may have undergone due to upstream exercise of privacy policy so as to allow robust computation of biomarkers that discriminate between privacy related data quality degradation and data missingness or degradation due causes such as battery exhaustion, network connectivity outages, sensor detachments, and others. As the data collected in CP7 studies will be publicly available and capture large amounts of sensor data in daily life, CP7 offers a tremendous opportunity for bi-directional interaction with TR&D3’s research under Aim 3. On the one hand, the privacy mechanisms developed under TR&D3 Aim 3 will provide CP7 with concrete instances of privacy related transformation that the metadata framework must capture. On the other hand, the metadata framework in CP7 and its stressors studies provides a vehicle for the TR&D3 Aim 3’s methods and tools for privacy-utility tradeoff in biomarker computation to be evaluated at scale and made available to the community in the form of concrete implementation. The interaction with CP7 regarding metadata mechanisms will be an iterative push-pull, whereby the feedback from deployment of the mechanisms in the studies will be used to refine the privacy mechanisms under Aim 3, which in turn will be used by CP7 to refine the metadata framework and re-evaluated in a new iteration of the Open Humans study.