Proposal

Narrative

PI (Dr. Ravi Shankar, Engineering and Computer Science), and Co-PI (Prof. Fran McAfee, Arts and Letters) have been collaborating since summer 2010 in developing smart phone Apps. Over 700 students from both these colleges as well as from the colleges of Design and Social Inquiry, Business, and Education, and at all levels (high school, undergraduate, and graduate),  have participated in courses offered by them in collaboration with faculty members in other colleges (Drs. Diana Mitsova-Boneva, Mate Titisawat,  and Alka Sapat from the college of Design and Social Inquiry; Dr. Michael Harris from Arts and Letters; Drs. Susanne Lapp and Don Ploger from Education; Dr. Ravi Behara from Business; and Ms. Agnes Nemeth from the A. D. Henderson School). Together, about 100 smart phone, semantic web, and social web apps have been developed. Recent ones are at these open source repositories: urban planning apps: https://bitbucket.org/shankarfau/profile/teams ; science museum exhibits apps: https://github.com/MODSApps ;  empowerment apps for middle school students (a research project that started out with apps developed in a class): https://github.com/EmpowerMe  ;  social web analytics apps: https://github.com/SocialWebApps ; and semantic web apps (for embedding domain expertise): https://github.com/Semantic-Web . Overview sites are listed here: http://android.fau.edu/ , http://robotics.fau.edu/ , and http://semanticweb.fau.edu/ .

We propose to extend our collaboration to address health care needs of the society today. Health care cost represents one-sixth of the entire US GDP (Gross Domestic Product) today. This is not a sustainable economic model that compromises everything else from the infrastructure to social security to education.  Concerted efforts have been underway for the past 10 years to reverse this trend, with leadership from the federal government during both Democratic and Republican presidencies.  ‘Meaningful Use’  rules from the Federal Government on the use and integration  of electronic health records (EHR) in moving from a prescriptive to preventive to predictive model of health care will lead to significant financial incentives to hospitals and clinicians; implementation of these is expected to improve patient’s qualify of life and life span, while reducing costs for the hospitals. A central theme is empowerment of the individual, whether a patient or a normal person, to take charge of their health and practice preventive and predictive measures to improve their health.

Personal Health Records (PHR) have evolved to address the needs of the patients, while maintaining privacy and confidentiality, as a distinct entity, different from Electronic Health Records (EHR) that are at the core of the ‘Meaningful Use’ directive.  PHR and EHR may be integrated in cases where the physician and hospital system support a holistic approach to health care. Under this vision, patients/normal individuals monitor their vital signs and symptoms using low cost biomedical, fitness, and environmental sensors at home and in the community; all such data is integrated with the clinical and lab data acquired during their annual physical exams and /or visits for their chronic conditions, if any. The physician, nurse, and social worker can coordinate care and advice to the person to help modify the risk factors to obtain better health outcomes. All this data can be continuously and regularly acquired; with patient’s permission, these can be used for for data analytics and big data insights (in the longer run) to develop best practices and expert rules.

We propose to build a lab resource with low touch/ non- touch sensors so a person can be monitored with as little interference, or disturbance from their daily lives, as possible. There are many low touch health sensors (for e.g., for temperature, blood pressure, heart rate, air flow rate, glucose measurements, etc.,) and environment sensors (for e.g., ozone, carbon monoxide, carbon dioxide, etc.,) that we plan to acquire. The non-touch sensors are exemplified by cameras that are useful for remote and automated monitoring. Recent products (based on computer vision algorithms) have identified breathing difficulties (as with sleep apnea and SIDS, or sudden infant death syndrome), suicidal tendencies (as with patients in mental hospitals and prisoners), and heart rate (from motion amplification of facial artery pulsations).  Cameras also support tele-health and counseling sessions. We will use cameras and VR (virtual reality) equipment to create avatars and recipes for proper exercising, with content input from nursing, physical therapy, and social work professionals. We are also acquiring Arduino kits to build platforms to interface such devices on a stand-alone basis. Arduino can then communicate with a gateway via a smart phone or a desktop PC. Home security companies such as ADT are working towards open source environments (see http://developers.ideallifeonline.com/ ). The model there is still subscription based; but we hope that is changing.  Our goal in building our open source infrastructure at FAU is to facilitate a cottage industry of entrepreneurs who can innovate and evolve different types of services, not merely a standard subscription-based system. That is, if the person is willing to take more responsibilities, perhaps with calls routed to their family or friends automatically, the new business models should allow that; this way, more people can get the health care support they need, even if they cannot pay for a full service. We would expect that Medicare/Medicaid or the insurance company would pay for installing a sensor network since that will reduce patient costs.

Another related goal is to build user and analytics apps to help people manage their health. Personal data will be collected, as described above, using sensors that are enabled to wirelessly connect with a gateway (a desktop PC, or a smart phone) that will then save the data in the cloud (or in a local storage). To this end we will enable low cost biomedical and environmental sensors for wired/wireless connectivity (some of them are enabled, while others are not – we will explore ways to enable all of them), develop user friendly apps,   integrate the acquired data (using a new standard called, FHIR (Fast Healthcare Interoperability Resources), from the HL7 organization), and analyze the data using social web and semantic web data analytics apps.  We have expertise in all of these areas.

The PI has background in embedded system design, electrical and computer engineering, data sciences, and app development. The Co-PI's strengths are complementary with a background in digital media, arts, animation, and user interface design (and several years of app development collaboration experience). This background will benefit the human interaction factor of the apps by making them intuitive and attractive. The growing trend to mix creative partners with the sciences empowers people to use technology to better their lives and not be impeded  by programming or coding expertise. Using avatars, virtual reality, and mixed reality techniques the courses will begin to develop experienced developers of these health related apps.

Our overall objective is to build resources for interfacing seamlessly different types of biomedical and environmental sensors to a common platform which can then communicate with a backend server via a smart phone or desktop PC. We expect to build the backend server with open source APIs of FHIR. Mirth Connect provides an open source and free solution, which we have downloaded and installed already. See:  https://www.mirth.com/Products-and-Services/Mirth-Connect?utm_source=google&utm_medium=cpc&utm_term=Download&utm_content=Mirth-Connect&utm_campaign=ADW_15_Mirth_Connect&gclid=CLXVzN-LjskCFY8dgQodIfQIvg . We will build a Web service RESTful client App that will run on the Android smart phone and Desktop PC to communicate with this server. The App will also interface with Arduino using Bluetooth. The infrastructure built will help us address issues of usability, user and sensor errors, alerts, trend analysis, data analytics, and eventually big data analytics. We will submit the protocol to IRB (Institutional Review Board) for approval. For this proposal implementation, we will collect data on students in the class, and with their permission, make it available for debugging, trouble-shooting, and analysis by other students, as feasible.  The PI is CITI certified in both medical and social sciences.

Relevance to the Strategic Plan: This proposal maps well to the following values: Teamwork (collaborations across disciplines by both faculty members and students); student success (students by undergoing team projects are better equipped for real-world team work; also the experience they gain will be state-of-the-art; further healthcare is an area that will need many more engineers and artists as  personalized health evolves); and Engagement (our work will directly help the community lead a better and healthier life style, in the longer run). This proposal addresses the following Pillars and Platforms: Healthy Aging; Sensing and Smart Systems; Big Data Analytics (eventually); Community Engagement and Economic Development (two of our teams from the past started businesses based on the app they developed in the class; they won FAU Student business competition - placing first and second, in two different years); Diversity (the app projects will bring diverse perspectives, especially when nursing and social students get involved; we hope that will happen in year 2, once we have decently integrated infrastructure); Global perspectives and participation; Leadership, Innovation, and Entrepreneurship (there is potential for many good ideas to come about, in the realm of sensor integration to a platform; two businesses were formed directly from our courses - see above; Please see the Resource Matching section also on the potential for small business formation); South Florida Culture (senior citizens make up a larger portion of our population; all our work can help them significantly); and Undergraduate Research and Inquiry (several of our undergraduate students have published papers in ASEE, the American Society of Engineering Education. This trend will continue with new student groups too).

Facilities

We do not need any renovations or upgrades. We have managed to use existing teaching lab resources by moving towards low cost, portable, modular, and reusable hardware and software components. We also upload all our apps to free open source repositories to share with the wider community. All these measures reduce cost and the need for a dedicated lab. We already have 20+ Nexus 7 2012/2013 tablets for our use in the courses, obtained with an earlier Tech Fee grant to the PI (three years ago). We have achieved much during the interim. Several conference publications have ensued. And of course, about 100 good apps are documented in open source sites such as Github, BitBucket, and Confluence.

Hardware Requirements

All the hardware and embedded systems to be acquired have the following characteristics:

1. We are focusing on low touch/ non touch sensors for patient monitoring. We have also included several environment sensors so we can learn about their usability and interfacing issues as well.

2. Cameras have been added to the list as a non touch sensor. Other researchers have used them not only for automatic monitoring of breathing abnormalities and suicide attempts (in mental hospitals and prisons), but also for more subtle applications such as measuring the heart rate (from motion amplification of arteries on the face). We will eventually explore such usages; however, initially, our focus in the courses would be to use the cameras to build a telehealth interface, and for physical therapy type of applications (to show how best to do an exercise).

3. Home health care also includes social and community support to help the person make conscious and concerted efforts towards modification. To this end, we wish to acquire units that will help us create Avatars and virtual reality experiences. We are preparing a NSF proposal in which we will work with professors in nursing (Dr. Teresa Sakraida) and social sciences (Dr. Naelys Luna) to ensure the content is relevant and helpful. We intend to team up students from nursing and social sciences with engineering and arts students to make sure the apps and products are on target. The previous two sentences indicate our plans for the NSF proposal. For this specific Tech Fee proposal, Drs.Sakraida and Luna may advise our engineering and arts students, as judges in project proposal and evaluation phases.

A detailed budget is provided under supporting documentation.

Software Requirements

Most of  the software to be used is open source; some of the sensors have embedded software that will be used as is. We have also developed much software in Java and Python for Android smart phone, semantic web, and social web data analytics apps. The backend server is an open source and free software from Mirth Connect, as explained in the narrative.

Personnel Costs

There are no requested personnel costs.

Other Costs

There are no other costs.

Timeline

Spring 2016 - We are scheduled to teach three courses that are relevant to the part on interfacing and developing smart phone apps: CEN 4214 - Software-Hardware CoDesign (to be offered by the PI); and COM 4930 - Designing Interfaces for Mobile Devices and DIG 6605 - Interactive Interface Design (to be offered by the Co-PI). All the courses have sufficient number of students enrolled already to form 10 to 12 groups of two students each from engineering and arts.  In these courses, the first half of the courses will be spent on discipline specific material. During the second half of the semester, the students will come together to interface one or two sensors per group to the Android smart phone and develop an user friendly app, using Java, XML, and Android Studio. We will invite nursing and social studies students/professors to help our students secure the proper perspectives to build these apps. Some of the sensors do not have a Bluetooth interface. If we cannot find a suitable interface, the students, if they choose that sensor,  will be tasked to connect them via the Arduino board, with the board communication via Bluetooth to the smart phone.

Spring 2016 - The PI has another course, a graduate course - COT 6930 - Web 2.0 Architectures and Algorithms. In this course, the students will focus on data analytics apps. The students will use the Mirthconnect FHIR server and develop Python code that will develop a RESTful client to run on desktops. Data from the undergraduate class, de-identified, and aggregated, will be made available for these students to explore. It is understood that the data may be limited in terms of their ability to draw meaningful analytics. However, the intent will be to find usability and user errors, as well as device limitations.

Summer 2016 - We plan to recruit DIS students to build the Arduino based platform to interface the sensors and aggregate them before sending them to a smart phone or desktop app.

Fall 2016 - we expect to offer the courses from Spring 2016 on smart phone app development. These courses are fairly popular and we have offered them at least twice in a given year. For this offering, we will focus on integrating the Arduino platform with smart phone apps. The sensors, in turn, will interface via Bluetooth or USB to the Arduino board. DIS students will be recruited to build apps for the desktop/smart phone Apps. We are able to recruit 2 to 3 DIS students at the undergraduate/ graduate level every semester. These are typically top students who are eager to achieve something beyond the class work on their own. Good results can typically be expected from such students. Not all the issues can be addressed in a given semester, however.

We will submit a NSF proposal on smart and connected health in December 2016.

2017 - We will use these courses to integrate better the platform with the Apps. The Apps themselves will have a data analytics component. All the work will be documented at open source Github repositories, so others can make use of our work. We expect to invite nursing and social science students to work with engineering and arts students to make the apps more realistic and relevant to the users.

2018 - We expect the work to be fairly mature by then. We will publish papers based on our experiences.

Sustainability

No additional funds are needed. We are securing low cost items. These typically last 5 years or so.

We will be able to manage with this one time funding. We also expect funding from other funding sources, especially NSF, to continue to explore this infrastructure for research purposes. 'Smart and Connected Health' is a major theme for funding within NSF, in collaboration with NIH. We believe that once this infrastructure is in place, it will be easy to seek such funding, hopefully by December 2016.

We are submitting NSF proposals based on this concept. Even if we do not succeed the first time (in December 2015), we will apply again, in Dec 2016; by that time, we will have a reasonably good prototype ready of our infrastructure, so that NSF is likely to fund us, in the area of smart and connected health.

The PI has brought in $4.5 M over the years in research funding, which is about 2 to 3 times the average amount in the college of engineering and computer science. His latest NSF proposal was ranked well and was recommended for 'funding if possible' ; however, it was not funded. We expect to resubmit and be funded. Usually it is a good sign if one's proposal was ranked so well and was recommended for funding. It is likely to convert into a funded proposal the following year. The PI has had 3 NSF grants in the past.

Resource Matching

The PI will make use of 20+ Nexus 2012/2013 tablets that he obtained 2 to 3 years ago with another Tech Fee grant. We have used them consistently through the many semesters when we have taught students App development.

We did not seek donations from any local company, such as ADT, because we want to develop an open source infrastructure; the sensors are not expensive and we have technical capability to build an open source infrastructure.

The consequence of the open source environment would be this: Some of our student groups can develop a service that is targeted to a specific disease community, such as for Asthma, Allergies, Alzheimer's Disease, etc. Each has unique requirements in terms of measurements to be made, and support and medication to be provided. Specialization will help them address one specific domain well and be successful as a business. Our open source infrastructure would have provide the backbone on which their service can be based on. Being the first ones to start their business in that domain, they are likely to be more successful than others who might follow them.

Implementing Organization

The CEECS Department of the College of Engineering and Computer Science will implement this. The PI belongs to this department. ?