Jog Falls: A Pervasive Healthcare Platform for Diabetes Management Lama Nachman1, Amit Baxi2, Sangeeta Bhattacharya2, Vivek Darera2, Piyush Deshpande2, Nagaraju Kodalapura2, Vincent Mageshkumar2, Satish Rath2, Junaith Shahabdeen1, Raviraja Acharya3 1Future Technology Research, Intel Labs, Santa Clara, CA 2Future Technology Research, Intel Labs, Bangalore, India 3Kasturba Medical College, Manipal University, Manipal, India Abstract. This paper presents Jog Falls, an end to end system to manage diabetes that blends activity and energy expenditure monitoring, diet-logging, and analysis of health data for patients and physicians. It describes the architectural details, sensing modalities, user interface and the physician���s backend portal. We show that the body wearable sensors accurately estimate the energy expenditure across a varied set of active and sedentary states through the fusion of heart rate and accelerometer data. The GUI ensures continuous engagement with the patient by showing the activity goals, current and past activity states and dietary records along with its nutritional values. The system also provides a comprehensive and unbiased view of the patient���s activity and food intake trends to the physician, hence increasing his/her effectiveness in coaching the patient. We conducted a user study using Jog Falls at Manipal University, a leading medical school in India. The study involved 15 participants, who used the system for 63 days. The results indicate a strong positive correlation between weight reduction and hours of use of the system. Keywords: Personal Health Monitoring, Diabetes Management, Energy Expenditure Analysis, Activity monitoring 1 Introduction Metabolic syndrome is emerging as a major public health issue across the world. It is a group of symptoms (e.g. central obesity, high blood pressure, insulin resistance) that increase the risk of heart disease, type-2 diabetes and stroke. Healthy life style characterized by increased physical activity and moderation of eating habits plays a key role in reducing these risk factors and slowing down the progression of type-2 diabetes or possibly even reversing it. To empower patients to better manage their disease, we envision a system that allows them to continuously monitor their physical activity and food intake, set goals and monitor progress towards these goals, allow them to reflect on their trends over an extended period of time and draw some actionable conclusions. Since physicians play an important role in coaching the patients, this system should be an effective tool

2 L Nachman, A Baxi, S Bhattacharya, V Darera, P Deshpande, N Kodalapura, V Mageshkumar, S Rath, J Shahabdeen, R Acharya in enabling physicians to be better coaches. Specifically we envision a backend portal giving physicians comprehensive and unbiased visibility into the patients��� life styles with respect to activity and food intake, as well as enabling them to track their progress towards agreed upon goals. To accomplish this vision, we worked closely with physicians to design a system for diabetes management which we call Jog Falls. In this paper we describe our Jog Falls system, its high level architecture and detailed design, highlighting the novel contributions in the different tiers of this architecture. Jog Falls is an integrated system for diabetes management providing the patients with continuous awareness of their diet and exercise, automatic capture of physical activity and energy expenditure, simple interface for food logging, ability to set and monitor goals and reflect on longer term trends. Its backend interface enables the physician to view the progress and compliance of the patients, hence facilitating personalized coaching. Finally its novel method for fusing Heart Rate (HR) and accelerometer data improves the accuracy of energy expenditure estimation, a key feature in enabling weight loss. We conducted a user study using Jog Falls at a leading medical school in India. India is on the verge of being the diabetes capital of the world [1], which calls for a comprehensive study in the Indian context. The study involved 15 participants, who used the system for a period of 63 days. We report the results of the study and discuss the effectiveness of this system in helping patients manage their lifestyles. Specifically the results indicate a strong positive correlation between weight reduction and hours of use of the system. The rest of the paper is organized as follows. We first present an overview of related work in section 2. We then describe the system design and implementation in section 3 and follow up with the details of our user study and the key learnings in section 4. We conclude with section 5. 2 Related Work Currently available solutions for diabetes management are discrete and disconnected. Activity and calorie expenditure monitoring devices such as pedometers quantify activity in terms of ���number of steps��� walked or calories expended. They use either spring based mechanical sensors [2], piezoelectrics [3] or accelerometers [4], [5] for activity monitoring. These devices are comparatively inexpensive and provide a reasonable solution for quantifying limited states like walking and running. However, they are not suitable for comprehensive activity tracking and are also susceptible to vibration errors. BodyBugg [6] uses additional sensors like heat flux, skin conductivity and temperature along with an accelerometer to improve the accuracy of calorie estimation. Several studies have positively evaluated the validity of BodyBugg in laboratory conditions [6]. The device is harnessed using an armband, which makes it a user friendly solution for long term monitoring. However, the lack of a user friendly way to track and set calorie goals is a major drawback of this system. Polar [7] uses Heart Rate (HR) to quantify activity in terms of calories, and is ideal for sports and fitness training applications. However, it is not designed for tracking continuous energy expenditure over extended time and doesn���t compensate for psychological factors that affect HR in sedentary conditions.

Jog Falls: A Pervasive Healthcare Platform for Diabetes Management 3 Moreover, this device uses a generic calibration curve (HR vs MET value) for all users, which reduces its accuracy. UbiFit [8] is a system developed to encourage individuals to self-monitor their physical activity and incorporate regular and varied activity into everyday life. The system components like glanceable display, interactive application, and a fitness device [9] provides valuable features like instant feedback on goals, long term log and journaling of activities and automated inference of several day to day activities. However, the system does not estimate calories burned while performing different activities. Other solutions like Houston [10], SHAKRA [11] and Fish���n���Steps [12] also suffer from the same limitation. Houston and Fish���n���Steps use pedometers to track step counts while SHAKRA uses cell phone signal strength variation patterns to track user activity. In Houston and SHAKRA, users can view real-time and historical step count/activity on the phone and also share this information with a group of friends. In Fish���n���Steps, user activity is presented as a game designed to encourage behavior change via social co-operation and competition. For dietary tracking, a major gap in today���s systems is keeping track of calories consumed in real-time. Websites like Nutrition Vista [13], Body Media [14], and FitDay [15] provide calorie intake estimates when the user enters the diet consumed, making them susceptible to recollection errors and hindering real-time self awareness. Myfoodphone [16] tracks food intake using photographs of the food captured by the users, thereby creating a photo-journal. The photos are sent using GSM/GPRS to a dietician, who assigns a goodness score based on preset targets. It also incorporates social facilitation, enabling the user to see progress of others. This system lacks real- time, automated feedback and requires regular intervention from the nutritionist. An iPhone app from Weight Watchers [18] features a large menu of food options, a point calculator that gives a score to users based on diet, and recipes that can be used to meet the point targets of the users. MAHI [17] is a solution to help newly diagnosed diabetics develop reflective thinking skills through social interaction with diabetes educators. It supports capture of interesting events that ���disrupt regular activities��� as opposed to capturing predefined activities like meals and exercise. While MAHI provides the user the flexibility to log experiences using voice notes and photographs as well as consult with diabetes educators, it does not support real-time activity and calorie expenditure/intake monitoring. It is clear that there is no solution that encompasses all the above elements in an integrated, simple and usable manner. Other features not readily available include enabling physicians to set goals remotely, regularly monitoring the patient���s condition, his/her adherence to food and activity goals, and coaching and motivating the patient to modify his/her lifestyle on a continuous basis. We tried to address these gaps by providing an integrated, usable and comprehensive framework that encompasses and manages all the above elements of diabetes management. 3 System Design and Implementation We defined the requirements of our Jog Falls system by working closely with physicians. The main goal of the system is to empower patients to manage their life