Addressing Food and Housing Insecurity among UCI Students

Community Engagement to Prevent and Address Food and Housing Insecurity among UCI Students through Smart Multimodal Automatic Lifelogging (UCI-FHI)

Food and housing insecurity are serious issues impacting academic success in university and college students. The number of students experiencing food insecurity and homelessness/housing insecurity, as well as the associated risk factors of emotional distress and substance use, is largely unknown among university students, including the University of California (UC). Findings of a 2015 system-wide UC survey revealed that 4 in 10 students across the 10 UC campuses reported not having a consistent source of quality nutritious food. Despite the creation of the UC Global Food Initiative which provides resources to many of the UC Campuses, universities continue to struggle to recognize, assess, and provide for the basic needs of vulnerable students enrolled in their systems. Thus, engaging frontline health workers, in

the form of trained peer student and community educators, along with realtime interaction via our low-cost technology-based monitoring system, will enable personalized screening of lifestyle and mental health of the University student population.  

Our solution is to pilot a comprehensive program at the UC Irvine campus to develop, assess, and model strategies preventing and addressing food and housing insecurity locally at the campus level, and to develop models which will have implications nationally and globally. Our solution leverages a holistic monitoring technology developed at UCI and combining emerging wearable technologies with a comprehensive monitoring system (multi-modal lifelogging framework) that students wear voluntarily to provide a realistic, time-oriented assessment of food and housing insecurity, emotional distress and substance use of students as well as analyze the root causes. Lifelogging is the process of extracting semantic level information from sensory data as well as user input to analyze people’s lives, and even environmental conditions and social situations. Our personal life-logging system automatically integrates sensory data from wearable devices (sleep, physiologically measured stress, physical activity), together with contextual, social, and environmental information from smartphones (location, social interactions) as well as moment-to-moment assessment using brief self-report questionnaires on an app. Our solution employs an event mining system to understand causal relationships among events, and learn relationships between them to build personalized models.

In this way, we can train our frontline health workers, led by faculty
and students of the Schools of Nursing, Medicine and Public Health, and our
community partners, to train additional peer student volunteers to predict
early behavior leading to negative consequences. Students currently inform us they feel embarrassed or stigmatized in reporting these issues to University staff, are unaware with whom to turn to, or lack accessibility in a timely manner, and may cope with the situation by taking drugs or alcohol.  Once we understand the full extent and the resources required, we will work together to design culturally-sensitive solutions to identify and address students’ needs. By garnering the resources of the university, and joining with innovative technological solutions and student-led initiatives and community partner resources, we will promote a comprehensive approach which will recognize the dimensions of the problem on each campus, and design solutions to fight these hardships early enough. 

We integrate community-based research with innovative technology to assess the problem and determine time-oriented solutions.  We work with campus faculty, staff and students, and public health and community organizations to comprehensively assess the problems which impact our student body. Contemporary life-event recognition technologies only capture low-level lifelogs (GPS, venue, or physical activity) and critically lack the ability to infer important factors that enable assessment of a person’s lifestyle. We offer a multi-modal personal life-logging system that automatically integrates sensory data from wearable devices together with contextual, social, and environmental information from smartphones  as well as app-based self-report questionnaires. 

We apply a novel lifelogging to build personal models to identify and direct students to resources that could reduce risks and build resilience. We classify students’ events into three main categories: Activity-related (e.g., “socializing”, “eating”, and “sleeping”), Mental Health-related (e.g., “stressful”, “happy”, and “depressed”), and Context-related (e.g., “car”, “home”, and “friend's home”). Then Situation Recognition is performed on higher abstracted data by recording events in chronological order. Certain situation, patterns and symptoms can then be detected, leveraging data analytic techniques. Our cloud-based data analytics and user interface system enables realtime interaction among frontline health workers and accelerates the decision support. 

The technology at the front-end consists of a smartphone-based lifelogger and the augmented version using an open-source smart wristband to collect more detailed information (e.g., physiological signs, stress, sleep, etc). Working with UCI Student Health Center, and engagement of frontline health workers, students will be encouraged to install the app so we can confidentially identify students who may be a potential or actual risk. These identified potentially food/home insecure students will be invited for interview, and the in-need group will be provided with the smart wristbands through which more comprehensive physiological and behavioral monitoring will be performed.

Our model will be scaled-up and made available to universities within the US and globally. After we develop the prototype at the UCI campus, over the subsequent 3 years, we plan to team up with the other UC campuses with a combined student body of over 250,000 students. Through forming a national alliance, we will then expand the reach of our solution across the US over the next 5 years. Due to the cloud-based back-end data analytics and app-based front-end data collection nature of our solution, its basic version (i.e., without smartbands) is scalable and requires minimal deployment costs.

Involving the student leadership of the campus will be critical in engaging the students in a survey to assess the prevalence of food and housing insecurity. Our focus groups with graduate students revealed that well trained students in each cohort tend to have good rapport with fellow students and can easily pick up on mental distress and substance use and may be the best first line of defense to recognize students in need. Engaging peer navigators in each cohort of academic program will encourage student participation in both survey completion as well as using the technology.

We have recently released our smartphone lifelogging engine through

Google Play Store and is under internal Beta test. We also setup a cloud server at UCI to serve as the back-end. Currently 10+ people who are participating with our solution. By August 2018, our user base will exceed 100 participants. For scalability, we plan to outsource the back-end system to cloud service providers such as Amazon AWS and Microsoft Azure. We will be the first to use this technology to identify and intervene with students at risk for food and housing insecurity as well as mental distress and substance use.

We plan to engage 1000 to install the app to identify potentially food/home insecure students. We are expecting to engage 50 students for the comprehensive monitoring phase using wearables. Via the app, the students receive alerts for resources immediately needed and are evaluated improvement of nutrition and sleep patterns, related to stable housing. With our community partners, we plan to garner enhancement of food pantries, temporary housing, and job skills and placement with reasonable work hours to ensure long-term stability. In 3 years, we are targeting full coverage of not only UCI students but also UC students in all campuses.

Computer sciences faculty: expertise in Internet-of-Things (Wearable Technology, Mobile Application, Cloud Services), Big Data Analytics (Event Mining, Context Recognition, Machine Learning), and Bio-signal Processing (Automatic Stress and Emotion Monitoring through Vital Signs)

School of Nursing, Medicine, and Social Ecology Faculty and staff: design the solutions and utilize existing university networks and data systems to collect baseline data, raise awareness, develop protocols, and activate internal university committees and networks

Public health and community partners: share community-based resources and strategies being developed and employed to address challenges for homeless people in Orange County

Student leaders: to inform the project and recruit participants

As a non-profit partnership, we intend to seek funding from public and private foundations to implement and deploy the technology and work together with university administration to integrate this model with the existing services offered by Student Centers and Basic Needs Units (e.g., http://www.basicneeds.uci.edu/). To date, this initiative has been developed through in-kind support from UCI faculty and staff leaders who have agreed to collaborate in addressing housing and food insecurity through community-wide efforts in Orange County to end homelessness and hunger. However, engaging key student bodies in the University, such as nursing, medical and public health students, peer frontline health workforce can be created to enhance referrals and coordination of care.

United Way is also interested in considering a partnership, as one of their goals is to reduce homelessness in Orange County. Finally, in addition to the $3.3 million initiative spent to expand the fight against campus malnutrition, we will tap UC President Janet Napolitano for possible support in this endeavor. The cost of the complementary package (smartbands) is also reasonable and can be covered through future partnerships with public and private organizations, and its integration to the rest of the system is feasible in a short period of time.

Solve will provide consultation to help us refine and grow this idea from a small, campus-based initiative into a statewide program.  Ultimately, we hope to scale up to offer a product that could be available to any campus both nationally and globally.  

By forming partnerships with other public and private organizations and investors via the extensive Solve network, we hope to receive feedback and constructive comments and solicit further support and receiving help to scale. Raising public awareness of our solution among the community, universities, foundations, and other institutions will also be important.

Student embarrassment to report food insecurity and housing issues is a major barrier. Thus, gaining the confidence, and support of the student leadership is so critical. Having the undergraduate and graduate student presidents onboard will allow the necessary channels of support so that trusting information will be passed onto the students. Strategies to promote privacy and confidentiality will be key and of high importance for the team.

Additional barriers will include the cost of the technology. If students are without a low-end smartphone, we will need to provide these smartphones to adequately address those students at risk.