Jiaqi (Jackey) Gong
      Home Research Teaching Publications CV      
In general, the research of my lab is focused on machine learning, data mining, systems and control optimization, and model-based embedded systems design and their applications in
1) Cyber-Physical-Social Systems design for monitoring, modeling and modifying human behavior, emotions, and brain.
2) Resilient Design Methodologies for Energy Harvesting Internet of Things.
3) Knowledge Discovery from unstructured, multiscale, incomplete and noisy data for health informatics, neuroscience, etc.

Active Grants:

2012-2016 NSF-SCH-EXP (PI: Maite Brandt-Pearce)
Personalized Signal Processing for Early Diagnosis of Mobility Impairment
Role: Key Personnel

This project leverages the general motivation of more personalized medicine to enable continuous, longitudinal gait assessments to be made using a Body Sensor Networks (BSN) for two example conditions with mobility impairment symptoms: normal pressure hydrocephalus (NPH) and multiple sclerosis (MS). Specific anticipated contributions include: 1) identifying signal features that can be feasibly extracted from out-of-clinic inertial BSN data and effectively utilized for detecting individualized gait changes, 2) developing new signal processing algorithms for the individualized identification and relative quantification of gait changes, and 3) exploring techniques for implementing forms of personalized signal processing on resource-constrained BSN platforms to enable more intelligent data reduction and dynamic energy optimization strategies that can extend the battery life of such systems.

Selected Publications:

[1] Jiaqi Gong, John Lach, Yanjun Qi and Myla D. Goldman, Causality Analysis of Inertial Body Sensors for Multiple Sclerosis Diagnostic Enhancement, Journal of Biomedical and Health Informatics, Vol. 20, No. 5, Sep. 2016
[2] Jiaqi Gong, Philip Asare, John Lach, and Yanjun Qi, Piecewise Linear Dynamical Model for Action Clustering from Real-World Deployments of Inertial Body Sensors, IEEE Transactions on Affective Computing, Vol. 7, No. 3, Sep. 2016
[3] Jiaqi Gong, Myla Goldman, and John Lach, DeepMotion: A Deep Convolutional Neural Network on Inertial Body Sensors for Gait Assessment in Multiple Sclerosis, Wireless Health Conference, Baltimore, MD, USA, Oct. 2016, In Press
[4] Jiaqi Gong, John Lach, Yanjun Qi and Myla D. Goldman, Causal Analysis of Inertial Body Sensors for Enhancing Gait Assessment Separability towards Multiple Sclerosis Diagnosis, 12th International Conference on Wearable and Implantable Body Sensor Networks (BSN), 2015 (Best Paper Award Finalist)


2012-2018 NSF-EEC (PI: Veena Misra)
NSF Nanosystems Engineering Research Center for Advanced Self-Powered Systems of Integrated Sensors and Technologies (ASSIST)
Role: Key Personnel, Techinical Coordinator

The center envisions a paradigm shift in health informatics enabled by battery-free wearable nanotechnologies that long-term monitor individual health parameters and environmental exposures. My research plays a key role in this vision to achieve proof-of-concept demonstration and human subject deployment of the testbeds through innovative top-down system approach that integrates incomplete knowledge from domain experts, heterogenous technologies from engineering groups and dynamical characteristics of the environmental parameters.

Selected Publications:

[1] Dawei Fan, Luis Lopez Ruiz, Jiaqi Gong*, and John Lach, “Profiling, Modeling, and Predicting Energy Harvesting for Self-Powered Body Sensor Platforms,” 13th International Conference on Wearable and Implantable Body Sensor Networks (BSN), 2016
[2] Jiaqi Gong and John Lach, “Reconfigurable Differential Accelerometer Platform for Inertial Body Sensor Networks,” IEEE Conference on Sensors, Baltimore, MD, USA, Nov. 2013 (Best Paper Finalist)

2014-2017 NIH-R01 (PI: David Peden)
Phase II Studies of Gamma Tocopherol As An Intervention For Environmental Asthma
Role: Key Personnel

In order to conduct phase IIa studies on determining if gamma tocopherol treatment of allergic asthmatics will prevent acute O3 and LPS-induced inflammation, my research is to develop lightweight personal wearable sensor arrays to provide real time continunous measures of O3, respiratory rate, ECG, ambient temperature, relative humidity, and personal energy expenditure (watts). These sensors will then be used in field studies to test the efficacy of either γT or inhaled corticosteroid for chemoprevention of O3-induced asthma exacerbation.

Selected Publications:

[1] Poster presentation in ASSIST center site visit in 2016. PDF

[2] Poster presentation in ASSIST center site visit in 2014.


2016-2019 NSF-CPS (PI: John Lach)
CPS: Synergy: Collaborative Research: Towards Dependable Self-Powered Things for the IoT
Role: Key Personnel

This project explores the fundemental and critically challenges caused by the mismatch between the characteristics of the energy harvesting systems and the physical world dynamics. Motion, temperature, light levels, wireless channel characteristics, and other physical world dynamics will be tracked in real-world, IoT-related deployments, such as wearables, critical infrastructure monitoring, smart buildings, and smart agriculture. These retrospective profiles will drive the design of adaptable energy harvesters (mechanical, thermoelectric, and solar), power management electronics, and control algorithms based in part on the prediction of future dynamics and their likely impact on node operation and energy harvesting. These fundamental contributions to and practical demonstrations of the science, technology, and engineering of CPS will lay the groundwork for enabling dependable self-powered wireless things in a massively-scaled IoT that provide highest fidelity performance given current and projected energy availability.

Selected Publications:

[1] Dawei Fan, Luis Lopez Ruiz, Jiaqi Gong*, and John Lach, “Profiling, Modeling, and Predicting Energy Harvesting for Self-Powered Body Sensor Platforms,” 13th International Conference on Wearable and Implantable Body Sensor Networks (BSN), 2016

2015-2019 NSF-SCH-INT (PI: John Stankovic)
Monitoring and Modeling Family Eating Dynamics (M2FED): Reducing Obesity without Focusing on Diet and Activity
Role: Key Personnel

Current behavioral science suggests that family eating dynamics (FED) have high potential to impact child and parent dietary intake and obesity rates. The confluence of technology research and behavioral science research creates the opportunity to change the focus of in situ obesity research and intervention from behaviors that have proven difficult to monitor, model, and modify (e.g., what and how much is being eaten) to the family mealtime and home food environment (e.g., who is eating, when, where, with whom, interpersonal stress), providing opportunities for monitoring and modeling (M2) behavior via remote sensing, and the potential for successful behavior modification via personalized, adaptable, real-time feedback.
This project proposes M2FED, an integrated system of in-home beacons, wireless and wearable sensors, and smartphones that collects synchronized real-time FED data that will be used to iteratively develop dynamic, contextualized FED systems models based on that data. The technology, ideographic models, and techniques to iteratively develop those models can guide future JITAIs and thus have a downstream positive impact on diet and ultimately obesity.

Selected Publications:

[1] Karen Rose, Yelena Perkhounkova, Jiaqi Gong, John Lach, John Stankovic, “Dignity in dementia: Implications for recruitment in to research studies”, Gerontological Society of America annual conference 2015
[2] Jiaqi Gong, Karen Rose, Ifat Emi, Janet Specht, Enamul Hoque, Dawei Fan, Sriram Dandu, Robert Dickson, Yelena Perkhounkova, John Lach, John Stankovic, “Home Wireless Sensing System for Monitoring Nighttime Agitation and Incontinence in Patients with Alzheimer’s Disease,” Wireless Health Conference, Bethesda, MD, USA, Oct. 2015
[3] Dawei Fan, Jiaqi Gong*, and John Lach, “Eating Gestures Detection by Tracking Finger Motion,” Wireless Health Conference, Baltimore, MD, USA, Oct. 2016, In Press

2015-2019 NSF-SCH-INT (PI: John Lach)
Behavioral and Environmental Sensing and Intervention for Dementia Caregiver Empowerment
Role: Key Personnel

Caregiver burden, stress, and depression resulting from person with dementia (PWD) agitation are the primary reasons cited for PWD transitioning from aging-in-place to a long-term care facility. Non-pharmacological interventions provided by caregivers can reduce the frequency and severity of agitation in PWD, but agitation can be unpredictable and influenced by the environment, and early signs of agitation often go undetected and can escalate to aggressive agitation that is more difficult to manage. As a result, most methods used to deal with agitation in dementia are reactive rather than proactive and are administered too late in an agitation escalation to be effective. A tool to predict agitation episodes and detect early stages of agitation would empower caregivers to intervene early and ultimately reduce agitation, thus reducing caregiver burden and extending aging-in-place and the associated quality-of-life and cost benefits.
This project will overcome the fundamental scientific barriers to realizing such a tool with a 3-phase research plan to employ Behavioral and Environmental Sensing and Intervention (BESI). First, a system of body-worn and in-home sensors will be developed to provide continuous, non-invasive agitation assessment and environmental context monitoring. Second, this system will be deployed in the homes of community-dwelling PWD to model the relationship between agitation and the environment and to identify features of early-stage agitation, both of which will inform the development of real-time caregiver notification strategies. Third, proof-of-concept for BESI will be demonstrated through a human subject pilot study, with caregivers receiving real-time notifications to alter the environment in advance of probable agitation episodes (based on trained agitation-environment relationship models) and to administer PWD interventions in the early stages of agitation (based on identified early-agitation features). Successful completion of this 3-phase project will motivate a follow-up large-scale proof-of-practice study.

Selected Publications:

[1] Jiaqi Gong, Philip Asare, John Lach, and Yanjun Qi, “Piecewise Linear Dynamical Model for Actions Clustering from Inertial Body Sensors with Considerations of Human Factors,” BodyNets: 9th International Conference on Body Area Networks, London, UK, Sep. 2014 (Best Paper Award)
[2] Jiaqi Gong, Sriram Raju Dandu, Bryson Reynolds, John Lach, and Jason Druzgal, “Unsupervised Head Impact Identification using Inertial Body Sensors based on Linear Dynamical Model,” Wireless Health Conference, Bethesda, MD, USA, Oct. 2014

2015-2017 Jeffress Trust (PI: Laura Barnes)
A New Computational Framework for the Prediction of Severe Sepsis and Response to Therapy
Role: Key Personnel

The goal of this work is to develop a new computational framework for the prediction of sepsis. This will include the identification of early sepsis indicators using commonly available electronic health records, clinical laboratory values, vital signs and predicting the personalized response to treatmenr post diagnosis.

Selected Publications:

[1] Jinghe Zhang, Haoyi Xiong, Jiaqi Gong, and Laura Barnes, “mGraph: Graph Representation for Electronic Health Record-Based Detection of Mental Health Disorders”, Journal of Biomedical and Health Informatics, 2016, Submitted
[2] Jinghe Zhang, Laura Barnes and Jiaqi Gong*, “Heterogeneous Convolutional Neural Networks for Electronic Health Records”, SIAM international Conference on Data Mining, 2017, Submitted.
[3] Jiaqi Gong, Jinghe Zhang, and Laura Barnes, “Risk Prediction using Electronic Health Records with Heterogeneous Graph Representation and Deep Convolutional Neural Networks”, Journal of the American Medical Informatics Association, 2016, In Preparation


Pending Grants:

2017-2021 IARPA-BAA-16-10 (PI: Northrop Grumman Systems Cop.) Submitted in Nov. 2016
Come Up with Something Creative for Multi-Modal Objective Sensing to Assess Individuals with Context
Role: Key Personnel at University of Virginia Site

This project is to discover the relationship between job context, individual difference variables and job performance through long-term monitoring and modeling of physiological and psychological measures captured by innovative mobile and wearable sensing systems.

Fig. Data representation and intepretation in two dimensions: multi-scale measures and individual difference variables.

Fig. Our pilot study collected multi-scale measures from smart phones of individuals with different anxiety levels. Strong associations between type of communication methods (text message and phone call) and the level of social anxiety is demonstrated.

2017-2020 NSF-SCH-INT (PI: Laura Barnes) Submitted in Dec. 2016
Multiscale Modeling and Intervention for Improving Long-Term Medication Adherence in Context
Role: Co-Principal Investigator

Figure. Conceptual representation of MMI. Left side shows the multiscale data streams captured by the MMI system consisting of wearables, smartphones, beacons, MEMS device, and SCT measures. Right side illustrates the multiscale modeling of the interconnections between the dynamic features extracted from the data streams (outside layer), the SCT contextual feature abstraction estimated from the EMAs and MEMS device data (immediate layer), and the abstraction of SCT model (inner layer). Deep learning architectures will be developed to explore the constructs underlying those interconnections. In addition, according to the construct information learned from the multiscale modeling, a model predictive control for intervention optimization will be implemented in two scales; micro-interventions such as text, prompts and video or audio clips (no more than 2 minutes) will be determined by rules considering of dynamic features and delivered just in time, while macro-interventions will be determined by the SCT contextual features.

2017-2019 NSF-SCH-EAGER (PI: Haiying Shen) Submitted in Dec. 2016
Optimizing Sensing Schedule in a Smart Home based on Personalized Medical Condition
Role: Co-Principal Investigator

The rapid adoption of remote health monitoring technologies across the U.S. healthcare systems coupled with the capability of linking perceptions of patients to healthcare providers in daily life provides a unique opportunity for conducting large-scale Personalized Medicine research. A critical step to make such research possible is the acceptability and feasibility of technology use in routine health care, from the perspectives of primary care clinicians, administrators, and clinic staff. A recent systematic review revealed aligning the output from smart home models with clinic workflow to be the most critical factor of the burden of the clinicians. This project develops optimizing sensing scheduling methodology to enable automatic alignment between the output of smart home models and the clinic data flow, which will empower primary care clinicians, administrators, and clinic staff.
The massive sensor data in smart homes are highly heterogeneous and multiscale, while the clinical data recorded by healthcare providers are sparse and heterogeneous as well. Although there are some existing algorithms for data alignment, they typically work with a single type of patient data and cannot handle those challenges mentioned above in general. This project develops adaptive sensing functionality of smart home by 1) learning the predictive model from the sensor data in smart home and the clinical information to determine the priorities of the sensors; and 2) adaptively schedule the sensing strategy in smart home towards automatically aligning the output of the smart home model to the clinic dataflow. Those functionalities will advance the intelligence of the physical systems, reduce burden clinicians experienced in reviewing the sensor data and ultimately maximize health impact in primary care.


2017-2019 Commonwealth Health Research Board (PI: Laura Barnes) Submitted in Sep. 2016
Optimized Predictive Monitoring of Non-Intensive Care Unit Patients at Risk for Sepsis
Role: Co-Principal Investigator

Update it soon!


In Preparation Grants:

2017-2019 R21: PAR-16-261 (PI: Jiaqi Gong) to submit in Feb.2017
EXCEL: EXamining and advanCing Emotional inteLligence of Healthcare Providers
Role: Principal Investigator

2017-2019 R21: PA-16-008 (PI: Jiaqi Gong) to submit in Feb.2017
Patient-Centered Monitoring, Modeling and Intervening Muscle Degeneration in Amyotrophic Lateral Sclerosis for Personalized Exercise Strategies
Role: Principal Investigator


Finished Grants:

2015 NSF-STTR Phase I (PI: John Lach)
Gait Tracker Shoe for Long Term Accurate Determination of Gait Parameters and Activity
Role: Key Personnel

2015-2016 ADE Medical Education Research and Innovation Grant (PI: Noah Schenkman)
Objective Motion Metrics for Surgical Skills Training
Role: Key Personnel

2011-2012 China Postdoctoral Science Foundation (PI: Jiaqi Gong)
Fast and Robust Image Point Sets Registration based on Probability Hypothesis Density and Gromov-Hausdorff Metric Theory
Role: Principal Investigator

Last updated in Dec. 14th, 2016
copyright @ Jiaqi Gong 2014