About Us: Overview
The mHealth Research Group at Northeastern University in Boston, MA is a transdisciplinary research team interested in using technology to radically improve health and wellness research and practice. We challenge ourselves to ask big questions:
- Ubiquitous sensing: Imagine that your smartwatch and smartphone can use unobtrusive sensing and statistical pattern recognition to infer what you are doing, as you are doing it. How might that transform healthcare as we know it, and how do we develop algorithms that can reliably measure everyday health-related behavior and habits and work 24 hours a day, seven days a week, 365 days a year? Specifically, how do humans and interfaces and machine learning algorithms all work together to create new opportunities for just-in-time adaptive interventions?
- Persuasive technology: Suppose we combine computers that detect behavior and context in real-time with our best health theories on human behavior change and decision-making. Can we develop powerful new theories and technologies to support just-in-time health-related behavior change systems?
- Personal health informatics: Soon our mobile and ubiquitous computing devices will continuously gather data on our health and health-related behaviors. How can new user-interfaces be built that provide person-facing personalized health systems that help individuals stay healthy and happy throughout their life?
This short "What if..." video summarizes a bit of what we want to do.
Our research team works with collaborators at Northeastern and internationally with the goal of answering such questions. We create new algorithms to detect behavior, new systems to change behavior, and new methods to accelerate research on health and wellness.
Our research team works with collaborators at Northeastern and internationally with the goal of answering such questions. We create new algorithms to detect behavior, new systems to change behavior, and new methods to accelerate research on health and wellness.
Our focus is on what is called personal, behavioral health informatics. Suppose a computer can acquire continuous, longitudinal data throughout everyday life as people use smartphones, smartwatches, miniature wearable sensors, and other in-home consumer electronics. How might the information gathered, and the new human-computer interaction opportunities created by the devices, be used to improve wellness?
Our research involves merging ideas from the computer science subfields of human-computer interaction, pattern recognition and machine learning, computational sensing, and artificial intelligence with ideas from behavioral science, behavioral medicine, social psychology, and preventive medicine. We are particularly interested in how algorithms that recognize everyday activities automatically or semi-automatically can drive the development of interactive preventive health tools that could ultimately be applied at the population scale in a cost-effective manner. Within computer science, this requires developing new user-in-the-loop activity detection algorithms that use context and common-sense information, without requiring large training sets. Within preventive medicine, this requires building and deploying pilot systems and demonstrating that the technology has a meaningful impact on health outcomes. To facilitate such work, our team has worked to create new tools that can be used to both measure and motivate behavior change using novel sensor-based technologies.
Our research involves merging ideas from the computer science subfields of human-computer interaction, pattern recognition and machine learning, computational sensing, and artificial intelligence with ideas from behavioral science, behavioral medicine, social psychology, and preventive medicine. We are particularly interested in how algorithms that recognize everyday activities automatically or semi-automatically can drive the development of interactive preventive health tools that could ultimately be applied at the population scale in a cost-effective manner. Within computer science, this requires developing new user-in-the-loop activity detection algorithms that use context and common-sense information, without requiring large training sets. Within preventive medicine, this requires building and deploying pilot systems and demonstrating that the technology has a meaningful impact on health outcomes. To facilitate such work, our team has worked to create new tools that can be used to both measure and motivate behavior change using novel sensor-based technologies.
Technology, Health, and team science
We strive for a truly transdisciplinary, team science approach. Rather than just working as a multidisciplinary team, where team members draw on knowledge from different disciplines but stay within their disciplines, in our work, all team members are encouraged to integrate health and technology knowledge to transcend traditional academic boundaries. This close collaboration is a distinctive attribute of nearly all of our work. We publish in both computer science venues, but also quite a bit in health venues.
We work on teams with collaborators throughout Northeastern University as well as at top medical research centers throughout the country. Current and past collaborating institutions include Stanford University, Duke University, the UT MD Anderson Cancer Center, the University of Southern California (USC), Case Western University, Temple University, and others. Projects have been funded by the National Institutes of Health, the National Science Foundation, the Robert Wood Johnson Foundation, industry, and Northeastern University.
The mHealth Research Group is based out of both the Khoury College of Computer Sciences and the Bouvé College of Health Sciences at Northeastern. Our team consists of students and staff who share a common goal of extending the capabilities of technology to dramatically improve healthcare as we know it today. It is this fusion of technology and health and behavior change theory upon which we focus our work, so that we innovate in both areas by leveraging advances in each.
We work on teams with collaborators throughout Northeastern University as well as at top medical research centers throughout the country. Current and past collaborating institutions include Stanford University, Duke University, the UT MD Anderson Cancer Center, the University of Southern California (USC), Case Western University, Temple University, and others. Projects have been funded by the National Institutes of Health, the National Science Foundation, the Robert Wood Johnson Foundation, industry, and Northeastern University.
The mHealth Research Group is based out of both the Khoury College of Computer Sciences and the Bouvé College of Health Sciences at Northeastern. Our team consists of students and staff who share a common goal of extending the capabilities of technology to dramatically improve healthcare as we know it today. It is this fusion of technology and health and behavior change theory upon which we focus our work, so that we innovate in both areas by leveraging advances in each.
Impact
Over the years, Prof. Intille and the mHealth group members have had an impact in five research areas.
Real-time wearable sensing and activity recognition
Since 2004, members of the group have been publishing work on the use of wearable sensors for automatic detection of everyday activity. This includes the development of systems that use single or multiple accelerometer data from wearable sensors and/or mobile phones to infer a person’s type, duration, intensity, and location of physical activity and sedentary behavior, in many cases in real-time. One focus has been on robustness and evaluation in field settings to support moving systems from the lab to the field for real-time for behavioral measurement and intervention in health domains, especially in physical activity research. Our group has developed wearable sensors and mobile phone algorithms for continuous, real-time wearable data collection on physical activity. That effort has led to additional, ongoing research on the recognition of stereotypies of autistic children (for research or educational system development), recognition of behavior in wheelchair users, detection of smoking behavior, and measurement of sleep quality.
Currently, we are most focused on how to integrate human-computer interaction with the algorithms, to create systems that work reliably 24/7 to detect physical activity, sedentary behavior, sleep, and many other health-related habits. In fact, our aim is to develop computer technology that will provide a comprehensive view of a person's behaviors, contexts, and internal states (e.g., mood).
Context-sensitive ecological momentary assessment (CS-EMA)
The group has extended its work on activity inference from sensors to propose the idea of context-sensitive ecological momentary assessment (CS-EMA), a research methodology that extends the increasingly popular technique of EMA to situations where a mobile phone queries for information (or delivers an intervention) in response to a specific automatically-detected behavior or context. Most recently, we have extended this work to explore what we call microinteraction EMA, or µEMA, which may allow temporally-dense data collection merging passive data gathering and frequent self-report.
Currently, we are most focused on how to use µEMA to realize our vision of a computer system with a comprehensive understanding of behavior and habits, and to demonstrate new research methods that might accelerate research in behavioral science and enable new just-in-time adaptive interfaces (JITAIs).
Innovative mobile health applications for measuring behavior
The group has demonstrated prototype systems that use technology in novel ways for health behavior measurement and intervention in domains including exercise science, weight loss, sleep, smoking, medication adherence, and autism.
Currently, we have projects looking at novel interventions for encouraging general health in communities of people (e.g., families/friend groups, church groups) and to encourage physical activity among wheelchair users with spinal cord injuries.
Fostering transdisciplinary (computer science + health) team science
Most broadly, the research group has contributed to bridging the fields of computer science and health behavior research, not only by working on transdisciplinary team-science projects, but also by organizing and participating in symposia and workshops to encourage others to do so as well. In addition, our lab is one of several that contribute to a unique PhD program at Northeastern in Personal Health Informatics, which is designed to foster students interested in team science in health and technology.
Currently, all of our projects are interdisciplinary. We work closely with colleagues in behavioral science and other health fields, which is both intellectually engaging and ensures that our research is grounded in solving real-world problems.
Home activity recognition and "living lab" research
Another line of work has been on detecting activities in the home using both wearable and environmental sensing, especially to support new research outside the lab (e.g., in homes). Contributions have included creating innovative examples of “living laboratories” such as the PlaceLab facility (an instrumented apartment with hundreds of embedded sensors) and creating sensors and algorithms that could be used to rapidly instrument typical environments for studying health-related behavior.
Currently, all of our projects are focused on mobile interfaces, because those are most practical for deployment in the field, but we are still intertsed in how to use mobile + environmental sensors, as the latter become more ubiquitous and accessible.
Our projects and publications pages provide more detail on our past research.
We are fortunate to work in a beautiful lab space in the heart of Boston. Read more about our workspace...
We are fortunate to work in a beautiful lab space in the heart of Boston. Read more about our workspace...
