Smart Mobile Health Systems
A key global challenge today is to deliver high quality yet economically efficient healthcare solutions. The prominence of mobile technologies holds the promise of fundamentally transforming today’s reactive, hospital-centered healthcare practice to proactive, individualized care, and shifting the focus from disease to wellbeing. While current mobile health literature is mainly focused on data collection and behavior monitoring, our methodology is to 1) build novel mobile systems that can monitor holistic health data ranging from biological rhythms (heart/respiration rate, sleep quality etc.), daily activities (work, exercise, eating etc.), to social interactions in dynamic, challenging environments, 2) transform health data to behavioral, psychological, and physiological models using novel sensing and data analytics techniques, and 3) develop motivational feedback systems to empower individuals to improve lifestyles and participate in their own health treatment.
This work is supported by a four-year NSF Smart and Connected Health (SCH) grant “SCH: INT: Collaborative Research: Unobtrusive Sensing and Motivational Feedback for Family Wellness”.
Guoliang Xing (PI, Associate Professor, MSU)
Tian Hao (Ph.D 2016, currently Research Scientist, Center for Computational Health, Watson Research Center, IBM) Chongguang Bi (Ph.D candidate, MSU)
Linlin Tu (Ph.D candidate, MSU)
Gang Zhou (Associate Professor, CS, Williams and Mary)
Jina Huh (Assistant Professor, Biomedical Informatics, UCSD)
Wei Peng (Associate Professor, Communications, MSU)
Barbara Smith (Professor, Nursing, MSU)
Awards & Recognitions
1. Best Mobile Application, third place, “iBreath: A Smartphone Application for Breathing Monitoring during Running” the 20th ACM Annual International Conference on Mobile Computing and Networking (MobiCom), Maui, Hawaii, 2014.
2. Startup Pitch Finalist, the 20th ACM Annual International Conference on Mobile Computing and Networking (MobiCom), Maui, Hawaii, 2014.3.
3. Best Mobile Application, third place, “iSleep: A Smartphone Application for Unobtrusive Sleep Quality Monitoring” the 19th Annual International Conference on Mobile Computing and Networking (MobiCom), Miami, FL, 2013.
Demos & Videos
iSleep – a mobile app for unobstrusive sleep monitoring (YouTube video)
iBreath – a mobile app for breath monitoring during running (YouTube video)
1. Chongguang Bi, Jun Huang, Guoliang Xing, Landu Jiang, Xue Liu, Minghua Chen, SafeWatch: A Wearable Hand Motion Tracking System for Improving Driving Safety, the 2nd IEEE International Conference on Internet-of-Things Design andImplementation (IoTDI), 2017, acceptance ratio: 15/51=29%.
2. Chongguang Bi, Guoliang Xing, Tian Hao, Jina Huh, Wei Peng, Mengyan Ma, FamilyLog: A Mobile System for Monitoring Family Mealtime Activities, the 15th IEEE International Conference on Pervasive Computing and Communications (PerCom), 2017.
3. Tian Hao, Guoliang Xing, Gang Zhou, RunBuddy: A Smartphone System for Running Rhythm Monitoring, The ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) 2015, acceptance ratio: 93/394 = 23.6%.
4. Tian Hao, Guoliang Xing, Gang Zhou, iSleep: Unobtrusive Sleep Quality Monitoring using Smartphones, The 11th ACM Conference on Embedded Networked Sensor Systems (SenSys), 2013, acceptance ratio: 21/123 = 17%.
5. Matt Keally, Gang Zhou, Guoliang Xing and Jianxun Wu, "Remora: Sensing resource sharing among smartphone-based body sensor networks," Quality of Service (IWQoS), 2013 IEEE/ACM 21st International Symposium on, Montreal, QC, 2013, pp. 1-10.
6. Matthew Keally, Gang Zhou, Guoliang Xing, Exploiting Sensing Diversity for Confident Sensing in Wireless Sensor Networks, the 30th IEEE International Conference on Computer Communications (INFOCOM), April 10-15, 2011, Shanghai, China, acceptance ratio: 291 / 1823 = 15.9%.
7. Matthew Keally, Gang Zhou, Guoliang Xing, Jianxin Wu, Andrew Pyles, PBN: Towards Practical Acitvity Recognition Using Smartphone-Based Body Sensor Networks, The 9th ACM Conference on Embedded Networked Sensor Systems (SenSys), Seattle, WA, Nov. 1-4, 2011, acceptance ratio: 24/123=19.5%.
1. Integrated mobile sensing and motivational feedback for family and community wellness. Going beyond individual health monitoring, we aim to develop integrated sensing and feedback systems that empower families and communities to actively engage in improving the wellness and tackling chronic diseases such as obesity, depression, diabetes, and dementia.
Research has shown that group routines such as family/social dinners, conversations, and group physical activity play a critical role in establishing good relationships among members and maintaining their physical and mental health. For instance, regularly eating dinner as a family reduces prevalence of child obesity and depression in the elderly. However, it is challenging to autonomously detect and log family/community routines for an extended period and in an unobtrusive, privacy-preserving manner. Moreover, while the current mobile health literature focuses on motivating individuals to pursue healthy life, systems that take a community-based approach have been unexplored. We are addressing these challenges through a new mobile health paradigm that integrates novel mobile sensing and motivation feedback systems. First, we proposed an accurate event classification approach based on ensemble learning. By exploiting the diversity in different sensor modalities, our approach significantly reduces reliance on user annotated ground truth. Second, we developed a new system for automatic detection of social mealtime activities using smartphones and smartwatches, including occurrence/duration of meal, conversations, participants, TV viewing, in an unobtrusive manner. This system is evaluated using 77 days of sensor data from 37 subjects in 8 families with children. We are extending this system to profile fined-grained social interactions within a community (school, workplace, and nursing homes). Third, based on the social comparison theory and HCI techniques, we are currently designing novel motivational feedback mechanisms, including virtual pets, visual metaphors, and social network tools, which improve self-awareness and engage the members in making continued progress toward improved routine.
Overview of our approach The architecture of family routine sensing system
2. Biological rhythm monitoring and regulation. A biological rhythm is any cyclic change in the level of a bodily chemical or function, including sleep/wakefulness cycle, body temperature, heartbeat, and respiration. They play a central role in maintaining our daily productivity and wellbeing. However, clinical monitoring of biological rhythms is often labor-intensive and cost-prohibitive. We are developing novel systems that leverage off-the-shelf smartphones and wearables for long-term, continuous, accurate biological rhythm monitoring and regulation. In [Ubicomp15], we proposed a new system for running rhythm (the coupling between breaths and strides) monitoring and regulation during exercise. As an important indicator for one’s fitness level and exercise efficiency, such coupling has been a focus of study in physiology. We developed a lightweight signal processing pipeline which detects breaths from the acoustic samples of Bluetooth microphone, and correlates strides detected through smartphone accelerometer. We employ a physiological model called Locomotor Respiratory Coupling (LRC) to mitigate the impact of significant noise from the environment (e.g., wind-induced microphone noise during running). Based on the LRC model, we are currently developing new HCI techniques to regulate an individual’s running rhythm, e.g., by using music tempos and vibration feedbacks, which will improve running efficiency and sustain beginning runners’ interest.
A typical setting of RunBuddy. The user is required to wear a Bluetooth headset and carry a smart-phone while running. RunBuddy comprises 3 major components: breath detection, stride detection and LRC-based calibration.
Screenshots of RunBuddy.
3. Unobtrusive sleep quality profiling. Over 40 million people in the United States suffer from long-term sleep disorders. The clinical polysomnograms-based sleep test may cost up to USD $5,000 per night. We developed a smartphone application called iSleep for unobtrusive, in-place sleep monitoring. iSleep samples the microphone of smartphone and employs new signal processing and machine learning algorithms to detect important events during sleep, including body movement, snoring and coughing. It then estimates the sleep cycle based on a clinically proven actigraphy model. The results can help users understand their sleep quality and detect early warning signs of diseases such as sleep apnea. A prototype of iSleep has been evaluated using 8 subjects and 51 nights of sleep. In collaboration with sleep physicians, we aim to go beyond the basic quantification of sleep quality to the provision of tailored behavioral feedback in the form of data-driven and clinician-approved actionable recommendations for improving sleep.
System architecture of iSleep
Illustration of sleep-related event detection and screenshots of iSleep app.